The epidemiology of antibiotic resistance: Policy levers
Download this section
Chapter 2

David L. Smith

The evolution of antibiotic resistance is defined as a change in the frequency of bacterial genes that affects the way bacteria populations respond to antibiotics, including their ability to grow at different drug concentrations and to persist through antibiotic chemotherapy. Ultimately, these genetic changes lower the effectiveness of antibiotic medicines and increase the length and cost of hospital stays, the probability of treatment failure, and the incidence of severe complications, such as chronic sequelae or death (Carbon 1999; Rubin, Harrington et al. 1999). This chapter focuses on the biological and epidemiological context for the evolution of antibiotic resistance and the policy levers that can be pulled to slow or possibly reverse the spread of resistance.

The biological study of antibiotic resistance spans many academic disciplines that are concerned with phenomena at, below, or above the level of a single bacterial cell. Cellular and molecular phenomena are relevant for finding new compounds to develop into drugs and for understanding the cellular and genetic mechanisms of resistance and cross resistance—that is, resistance to several drugs that is conferred by a single genetic mechanism (Box 2.1). These phenomena set the stage for understanding the evolution of resistance, but cellular and molecular phenomena are only distantly related to risk and cannot be manipulated through policy; it is hard to imagine a policy that would target individual cells, except for chemotherapy itself.

Our focus is on those aspects of the biology that are directly related to changes in risk and can be affected through policy—typically phenomena above the level of a cell, such as population dynamics and control. The evolution of resistance and the associated increased risks of treatment failure involve not just the mechanisms that confer resistance to antibiotics, but also the likelihood that some bacteria are resistant to chemotherapy, regardless of the biological mechanism. The cell-level conceptual divide does not always cleanly delimit the relevant subjects; for example, the movement of plasmids among bacteria is a subcellular phenomenon with population-level consequences. The important distinction is whether the biological process is related to changing risk within some population of interest.

The approach here is eclectic; we borrow from many academic disciplines that are concerned with population-level phenomena, including ecology, microbial ecology, population biology, infectious disease epidemiology, evolutionary biology, and population genetics. Changes in risk are quantitative phenomena, so the study of antibiotic resistance inevitably involves some sort of mathematical model. Mathematical models of the spread of antibiotic resistance help synthesize information from multiple disciplines, and they are important conceptual tools for evaluating control measures (Box 2.2). This chapter begins by describing the biological background for the problem and then proceeds to the policy levers that can be pulled to change the biological world.

The concepts that are important for understanding the evolution of antibiotic resistance and its control are the ecology and structure of bacteria populations, the epidemiology of infection, the treatment of an infection, and the origins and spread of resistant bacteria. This chapter discusses each of these concepts in turn. We then examine the goals of controlling resistance and define an appropriate aim—to maximize the total net benefit of antibiotics in a population over time. Once the goal is articulated, we can look at strategies for achieving it: controlling antibiotic use, controlling infections, and pursuing interesting new approaches based in ecology. Finally, we review some success stories and identify common threads.

Structure of bacteria populations

To describe changes in the frequency of resistance within a population, it is first necessary to define a population, but this is not an easy task. Bacteria populations and communities have a complex distribution and functional structure. One natural definition of a bacteria population is all the bacteria of a single species within a single host or part of a host. Another is all the bacteria that are shared among a well-defined population of hosts. Both definitions are relevant for the epidemiology of resistance. Within any single human, bacteria populations come and go, but even as the local population is subject to turnover, the bacteria population in a collection of hosts is stable because of transmission or colonization from other hosts. In ecology, structured populations that are characterized by local demographic instability but global stability through recolonization are called metapopulations.

The emergence of resistant bacteria within a single host and the spread of resistant bacteria or genes (McGowan 1983; Davies 1994) among hosts are closely interrelated processes, and substantial research has been directed at understanding the interplay between the processes that occur within and among hosts (Levin 2001; Lipsitch 2001). Most host populations, moreover, have their own spatial distribution and contact network, and this imposes a higher-order structure on bacteria populations (Smith, Dushoff et al. 2004). Bacteria can also have an important reservoir outside, in a nonhuman host or the environment (Kummerer 2004). Since bacteria spread within and among hosts and even shift host species, the proper definition of a bacteria population is determined by the question at hand and by the temporal and spatial scale that it implies.

For bacteria, humans are one kind of habitat. The bacteria populations within a single human host have a particular niche, a mode of life in some part of the body where they are typically found in healthy people. For many bacteria species, the gut is the primary habitat. Other important microbial communities in humans are found in the mouth and nose, the skin, the ear, and the vaginal tract. Most of the time, the bacteria populations in these habitats are harmless to their host, and some are beneficial. Bacteria populations that persist on a host asymptomatically are said to colonize a host, and the host is said to be a carrier. Colonization with commensal bacteria, even resistant bacteria, is not generally regarded as an important medical phenomenon per se. Colonization with bacteria is a natural, necessary, and inevitable condition of life.

Bacteria, including resistant bacteria, become a medical problem when they cause an infection—the invasion of tissues where bacteria are not normally found, such as the bloodstream, lungs, urinary tract, other sterile sites of the body, or wounds. For example, enterococci are commonly found in the gut, but they can cause a life-threatening infection when they enter the bloodstream, often through a wound, and begin to proliferate. Infection is generally accompanied by some symptoms, but the line between colonization and infection is sometimes fuzzy.

Colonizing bacteria compete with other bacteria of the same species and of different species for the habitat they occupy. The bacteria communities that inhabit an individual human are some mix of stable flora, bacteria that persist in the host for years, and transient flora—bacteria that were recently acquired and that are just passing through. The stability of bacterial flora may be due to persistence in refugia inside the body—surfaces to which they may adhere or that otherwise protect them from other microbes or the immune system.

Many factors affect the composition of the stable bacterial flora that inhabit a host. Bacteria colonize the body within the first few days of life. Most bacteria produce some metabolic by-products that alter the environment; these chemicals may help construct a niche in their local environment that favors them, a niche that could last throughout the life of the host. The immune system and physical environment combine to shape the bacterial flora. Immune chemicals act as a sort of top-down regulation that influences the composition of the microbial community. Similar principles apply to the finegrain-structure body surfaces where bacteria adhere.

Bacteria compete with one another indirectly by consuming scarce resources, and directly through a sort of chemical warfare, or by changing the chemical composition of the surrounding environment (Dykes 1995; Jack, Tagg et al. 1995; Dykes and Hastings 1997; Riley and Wertz 2002). On top of the interactions with other bacteria, bacteria may be subject to infections themselves. Some viruses, called bacteriophages, invade bacteria populations sporadically and limit or eliminate their populations. Infections by other parasites can change the community indirectly, mediated through an immune response that changes the flora or through antimicrobial use. Most of these processes are poorly studied but contribute to the enormous variability in bacterial flora over time, including sporadic changes to the “stable” flora.

The perturbations caused by antibiotic use are the most important factor in the evolution of antibiotic resistance. Antibiotics eliminate sensitive bacteria and open up niches for resistant bacteria or other species to grow, and they can also cause wholesale changes to the gut, such as antibiotic-associated diarrhea. Another effect is bacterial overgrowth—a rapid increase in the bacterial load of some species, a side effect of eliminating other species with antibiotics (Donskey, Hanrahan et al. 1999; Donskey, Chowdhry et al. 2000; Donskey, Hanrahan et al. 2000). During these periods of perturbation, the bacterial clones that make up the transient flora may expand and find a more permanent place within the gut. Such events may be important for helping antibiotic-resistant bacteria become a part of the stable flora, thereby making the host a carrier.

An important but often overlooked principle is that antibiotic use selects for resistance in several different species of bacteria simultaneously, regardless of the reasons why a patient takes the drug. When a patient takes an antibiotic to treat an infection, the concentrations of that antibiotic become elevated and affect target as well as non-target species, including bacteria from many microhabitats throughout the body. The mode of administering the antibiotic does have some effect because it affects the concentrations of the drug in other parts of the body; antibiotics typically reach much higher concentrations in the gut if given orally rather than intravenously (Drusano 2005).

All this suggests that the human body is a highly dynamic and variable environment. Given this unpredictability, no host is likely to provide a stable home forever. Thus, bacteria’s mode of life within a host is only one component of bacterial fitness. To persist within a population of hosts over time, bacteria must spread to other hosts. The spread of bacteria (and resistant bacteria) that stabilizes bacteria metapopulations is a largely invisible process involving colonization of healthy humans, with occasional infection. A human population represents a bacteria metapopulation with a particular kind of structure (i.e., the spatial distribution and contact network of the hosts) that affects the way bacteria spread.

Epidemiologically important contact occurs as a part of normal life. Opportunities for spread are more common among people who are frequently in contact, including family members, coworkers, schoolmates, or members of a church or health club. Other important types of structure pitals, where health care workers act as vectors, and more structure is added by the exchange of patients among hospitals, long-term care facilities, and the surrounding community (Trick, Kuehnert et al. 1999; Smith, Dushoff et al. 2004). Also associated with the spread of antibiotic-resistant bacteria are intravenous drug use (Saravolatz, Markowitz et al. 1982; Saravolatz, Pohlod et al. 1982) and prison populations (Aiello, Lowy et al. 2006). Typically, bacteria spread to those people with whom contact occurs most often, but eventually, bacteria can spread among less connected populations. At regional, national, and global levels, resistant bacteria spread through international business and travel (Okeke and Edelman 2001; O’Brien 2002). The structures of human population that determine contact work together with the natural changes in bacterial flora within a body, thus providing opportunity for bacteria (including resistant bacteria) to spread among populations.

The complement of transmission is persistence: the among-host component of bacterial fitness is the product of the rate at which bacteria are transmitted to other hosts per day and the number of days that the bacteria persist. In other words, bacteria can increase their among-host component of fitness by spreading more efficiently, or by persisting in a host and “shedding”1 for a longer period of time. One of the important fitness effects of antibiotics is to shorten the persistence times of sensitive bacteria by depressing their among-host fitness relative to antibiotic-resistant bacteria. In some cases, colonization with antibiotic-resistant bacteria is extremely persistent; a carrier can shed the same bacteria for years (Bonten, Hayden et al. 1996; Henning, Delencastre et 6). In fact  carriers have been shown to play a special role in epidemics (Smith, Dushoff et al. 2004). The efficiency of control might be vastly improved by finding and targeting carriers—by isolation during hospitalization or selective decontamination (see discussion below).

From a medical-ecological perspective, bacteria differ in many ways: their niche, their propensity to colonize and infect humans, their effect on human health, their ability to spread among humans, and the severity of symptoms when they cause disease. The antibiotic resistance problem in the United States is split between hospital-acquired and community-acquired pathogens; most of the bacteria are hospital acquired except for Streptococcus pneumonia (a.k.a. pneumococci) and a recently emerged community-acquired Staphylococcus aureus. Antibiotic resistance is common in both Gram-negative bacteria and Gram-positive bacteria, such as staphylococci, streptococci, and enterococci.

Epidemiology of infection

When bacteria infect an ordinarily sterile site, they present a serious medical condition, even if they are not resistant to antibiotics. To understand bacterial infections, it is important to characterize the source: where were the progenitors of the infecting bacteria a few generations previously? Here, we give an  pidemiological, ecological, non-clinical overview of bacterial infections and the ecological reservoir.

Perhaps the most important ecological reservoir for an infection is the population of bacteria that colonize some other part of the host’s body. The most common route of infection is the spread of bacteria from one part of a host’s body to another, moved around by the host itself or by a caregiver. In many cases, the infection starts from contamination with colonizing flora during a medical procedure.

In other cases, the bacterial infection is spread from other hosts, including other hospital patients, hospital workers, family members, or schoolmates (Bonten, Slaughter et al. 1998). The relative importance of these host reservoirs as a source of infection probably declines as a function of proximity to the focal host; the most likely sources are the patients themselves, followed by health care workers, other hospital patients, and family members.

The bacteria are transmitted by direct contact, such as touching or sneezing, or indirect contact through an intermediate contaminated object (a “fomite”). For example, health care workers can be carriers, or they may be vectors who move bacteria among patients or from contaminated objects in a patient’s room. The objects that surround individuals, including furniture and food and water, can become contaminated. Medical devices are a particularly important source of infections (Lund, Agvald-Ohman et al. 2002; Agvald-Ohman, Lund et al. 2004): they bring a potentially contaminated surface into contact with living tissue. One problem with medical devices is that their wet surfaces facilitate the growth of biofilms (Box 2.3), which can help facilitate gene exchange and persistence, protect bacteria from antibiotics, and so provide a natural refuge and gentle exposure that may become important in the evolution of resistance (Costerton, Stewart et al. 1999).

A potential source of antibiotic resistance in environmental bacteria is the sewage effluent from hospitals and long-term care facilities, which contains large numbers of resistant bacteria (Kummerer 2004). Large amounts of antibiotics are also used in agriculture for prophylaxis or as nutritional supplements, and antibiotic-resistant bacteria can remain in meat through the abattoir and retail (Witte 1998). Most meat is properly cooked in the home or in restaurants, but uncooked meat can cross-contaminate raw foods during preparation. This is a potentially important source of exposure and perhaps colonization. Like hospital sewage, the effluent from farms that use antibiotics can be a source of antibiotic-resistant bacteria in the environment (Witte 1998). Alternatively, farmers may become colonized by novel strains of antibiotic-resistant bacteria and transmit them into the population (Aubry-Damon, Grenet et al. 2004). Other sources of bacterial infections are the animals themselves. Infections with zoonotic pathogens are typically acquired from contact with animal food products or animals, usually without subsequent transmission to other humans (Box 2.4).

The human body is constantly bombarded by bacteria, but most potential bacterial infections are prevented by the immune system (Levin and Antia 2001). Infections often begin when the immune system is compromised. The skin provides the first and most important protection, but wounds compromise immune protection and allow bacteria to gain access to blood and other tissues. The risk of infection is further exacerbated when medical devices contaminated with biofilms bring the bacterial world into close contact with ordinarily sterile sites, especially the insertion of intravenous needles and tubes to aid with breathing or urination.

In patients with compromised immune systems, including the elderly, patients with HIV/AIDS, cancer or transplant patients, and patients who have recently had influenza, bacterial infections have become increasingly common. Such patients constitute an important segment of the population at risk from antibiotic-resistant infections. Their compromised immune status and multiple health problems make computing the burden of disease for antibiotic resistance difficult.

Bacterial infections grow exponentially at first, so a small initial population of bacteria cells can threaten a patient’s life within a few days. The longer a bacterial infection continues without treatment, the higher the peak bacteria population densities, the greater the genetic diversity, and the greater the risk of complications (Paterson and Rice 2003). Left untreated, acute bacterial infections can develop into chronic infections when bacteria adhere to body tissues (Fux, Stoodley et al. 2003).

Treatment of infection

Not all infections need antibiotics, not all are curable by antibiotics, and not all treatment failures are due to evolved resistance. Once an infection begins, the immune system mounts a response that limits or clears an infection. Antibiotic treatment, however, substantially limits the duration and severity of the infection, especially when the immune response is insufficient. Antibiotic therapy often shortens the duration of symptoms and decreases the likelihood of complications and death.

All else equal, the earlier an effective antibiotic is given during the course of an infection, the better the prognosis for a successful recovery. Since infections grow exponentially, at least initially, the earlier antibiotics are used, the less work they have to do. When a patient presents with symptoms that resemble a bacterial infection, a doctor who suspects a bacterial infection on the basis of her initial diagnosis usually chooses an antibiotic without a microbiological confirmation of the infecting agent. This method for selecting antibiotics, called empiric therapy, is common practice. Since antibiotics are most effective if they are used immediately, empiric therapy is preferred, and it is probably good for patients (Paterson and Rice 2003). On the other hand, the risk of an adverse outcome differs considerably, depending on the primary diagnosis. When intervention is not urgent, some delays in prescribing an antibiotic may not increase the risk to the patient and might substantially decrease the use of antibiotics (Edwards, Dennison et al. 2003).

Unfortunately, without a confirmed diagnosis, the doctor must choose among several antibiotics and logically goes with her best bet—a broad-spectrum antibiotic, one that covers the broadest possible range of infecting agents, rather than an antibiotic that targets only a few. Often patients continue a course of antibiotics even after a microbiological test fails to find a bacterial infection. If a bacterial infection is confirmed and the patient is responding to treatment with the initial antibiotic, the doctor is generally reluctant to switch to a narrow-spectrum drug that might have been more appropriate in the first place. (Chapter 3 explores the issue of overprescribing in greater depth.)

A specimen is sometimes sent to the clinic’s microbiology laboratory. Lab work becomes extremely valuable if a patient fails to respond to treatment, but treatment failure is uncommon and lab work has a tangible cost. If a microbiological sample was not taken initially, further waiting is required, and this can put patients at further risk.

A microbiological lab report about an antibiotic generally includes the species of bacteria that are present in an infection and a profile of the resistance patterns. These microbiological patterns also form the basis for much of the existing surveillance data. Importantly, the definition of antibiotic resistance from microbiological tests (i.e., in vitro) does not always correspond to the results of treatment (i.e., in vivo). These differences affect the ability of doctors to appropriately treat patients and may introduce a major source of bias in all the passive reporting on the frequency of resistance.

Antibiotic therapy can fail for reasons that have nothing to do with evolved resistance. Many bacteria species are intrinsically resistant to antibiotics: some antibiotics are effective against only Gram-positive or only Gram-negative bacteria, some work only on certain species. Bacteria can acquire phenotypic resistance, a kind of acquired resistance that arises in response to antibiotics and is not inherited (Levin 2004; Wiuff, Zappala et al. 2005). Bacteria in biofilms (Box 2.3), which form on medical devices and human tissues, can become persistent sources of infection that are not easy to eliminate or cure (Rotun, McMath et al. 1999). Thus, the evolution of resistance is one of many causes of treatment failure.

Origins of resistant bacteria

Evolved resistance is of greatest interest to policymakers because it is responsible for the increase in the frequency of resistance and because it holds the promise of being managed through policy mechanisms. In practice, policy approaches have met with varying degrees of success. Some bacteria evolve resistance to antibiotics immediately; other species wait decades before resistance emerges, and the reasons for this difference are not well understood. Here, we consider origins and spread as different and important steps in the evolution of resistance. First, we consider the origins of resistance, because the genetic origins of resistance may affect the choice of policy (for a longer discussion of this subject, see Lipsitch and Samore 2002).

Resistant bacteria can evolve de novo from a population that was sensitive before treatment in two ways: quantitative changes in resistance through the accumulation of random mutations, and the transfer of whole genes or sets of genes from other bacteria species (Davies 1994). In the first way, populations can evolve gradually and quantitatively through simple point mutations, which are one-letter changes in the genetic code, or through more significant mutations, such as insertions or deletions in several letters that are mistakenly omitted or copied from somewhere else. This is how other pathogens commonly evolve resistance. In the second way, the potential to acquire whole sets of genes from other species distinguishes bacteria from viruses or eukaryotic pathogens.2 Either way, for bacteria to become viable and epidemiologically important, further genetic changes may be required after transmission among several hosts and repeated exposure to antibiotics. The evolution of resistance de novo refers to the process of selection within a host that transforms a population of sensitive bacteria with a few resistant mutants into a daughter population of bacteria that are efficiently transmitted and dominated by resistant mutants.

Incremental changes in resistance have been understood since antibiotic resistance was discovered. Large phenotypic changes from small genetic changes are possible, but point mutations typically provide small advantages by themselves. Thus, bacteria must accumulate many small mutations to tolerate the high concentrations of drugs given to patients. Steady, incremental increases over time in bacteria’s ability to tolerate drugs can be countered by steady increases in the dose of a drug until the toxicity of the drug at high concentrations becomes a problem. If patients do not comply with a drug regimen, some bacteria survive that are more resistant. These partially sensitive bacteria spread and mutate, and resistance can increase on further exposure to antibiotics. The solution has been to ensure that patients take a sufficiently massive dose of a drug to completely eliminate the pathogen. Thus, to discourage the rapid evolution of quantitative resistance, emphasis has been placed on compliance with antibiotic treatment, and this has led to the perspective that the evolution of resistance in bacteria is largely an issue of the abuse of antibiotics, including patients’ noncompliance and physicians’ overprescription.

That point of view, at least with respect to noncompliance with antibiotic drug regimens and the origins of resistance, is not entirely correct, since resistance can emerge by the acquisition of whole genes on genetic elements that are transferred among bacteria, including multiple-drug resistance that is transferred all at once. Many bacteria share genetic material with other closely related and sometimes more distantly related bacteria. In this way, whole genes or sets of genes that work together can move out of one ecological reservoir, such as a soil microbe or farm animal communities, and into another, such as hospital-acquired pathogens. This has been an important way for antibiotic resistance to develop and spread, especially in the case of vancomycin-resistant enterococci (VRE) and methicillinresistant S. aureus (MRSA) and pneumococci (Lipsitch and Samore 2002). Thus, the spread of resistance genes, and not necessarily of bacteria, can be one underlying cause of an epidemic of resistance. Moreover, genes that confer resistance to antibiotics can move around together to provide bacteria with an easy way of becoming resistant to many antibiotics all at once.

Implementing a policy of increasing compliance to discourage the origins of resistance may or may not delay the emergence of resistance, depending on the underlying genetic mechanism. For resistance that is transmitted on mobile genetic elements, compliance is not likely to be effective in delaying the emergence of resistance, and the increased length of compliant treatment prolongs the period of increased risk of colonization by resistant bacteria. In some cases, increased compliance may be counterproductive (Lipsitch and Samore 2002).

Spread of resistant bacteria

Despite the interest in the emergence of antibiotic resistance, the problem is largely the spread of resistant bacteria or resistance genes, not the repeated evolution of resistance de novo (Box 2.5, see next page). Indeed, the spread of a resistant bacterial clone or resistance genes is probably responsible for more clinical failure than de novo resistance within a host, so controlling spread should be given much greater concern and attention. In fact, policy should be chosen based on the propensity for resistant bacteria to evolve de novo in response to antibiotic use, or to spread. These propensities differ among bacteria species, so different policy solutions may be required for different bacteria.

Why do antibiotic-resistant bacteria spread? This question is, perhaps, wrong minded. Since all bacteria spread, given the opportunity, why would antibiotic-resistant bacteria not spread? Not all bacteria are successful; ultimately, bacteria are limited by many other factors, including other bacteria. In the intensely competitive microbial world, resistant bacteria are thought to be at an inherent disadvantage because of the biological cost of resistance (Box 2.6). The weight of evidence supporting a biological cost of resistance includes some strong a priori reasoning, in addition to direct observational and experimental evidence (Andersson and Levin 1999). The biological costs may be manifest as slower growth rates within a host that would make resistant bacteria lose out in head-tohead competition with other bacteria. Resistant bacteria may be shed at lower rates and spread less efficiently, or they may clear faster, thus providing fewer opportunities to spread. Evidence also suggests that the biological cost of resistance is highest just after resistance evolves, and then declines over time as compensatory mutations arise that reduce the biological cost of resistance (Levin, Perrot et al. 2000).

Countering the biological cost of resistance, the use of antibiotics in a bacteria population confers two main fitness advantages to the resistant bacteria. First, the use of antibiotics reduces the fitness of sensitive bacteria by increasing their clearance rates. Second, as the concentrations of antibiotic wane, resistant bacteria have a window of opportunity when they can colonize a host but sensitive bacteria cannot. When these fitness advantages are strong enough to compensate for the biological cost of resistance (see Box 2.6), resistant bacteria tend to spread.

Put another way, the use of antibiotics shifts the balance in favor of resistant bacteria over sensitive bacteria, and the among-host fitness advantages created by antibiotic use outweigh the biological cost of resistance. When these ideas are encapsulated into a mathematical model, a simple principle arises: there are thresholds on the rate of antibiotic use that favor resistant bacteria (Austin, Kristinsson et al. 1999). If the rate of use is below the threshold, the frequency of resistance tends to decline. If the rate of use is high enough, resistant bacteria increase in frequency. Thus, an important factor is the rate that an antibiotic is used in a population.

The rate of antibiotic use is too low in most places to favor resistance, but antibiotics are heavily used in hospitals and other health care facilities. These institutions are connected by repeated hospitalization and the transfer of patients to and from hospitals and long-term care facilities (Smith, Dushoff et al. 2004). This leads to a view of the resistance epidemic that focuses on controlling transmission of resistant bacteria in places where antibiotic use is heaviest (HICPAC 1995; Weinstein 2001). Hospitals and long-term care facilities (Nicolle 2001; Elizaga, Weinstein et al. 2002) are regarded as “sources” because a greater fraction of patients leave a hospital colonized with resistant bacteria than entered, and the surrounding community is a “sink” where the frequency of colonization tends to decrease. The declining resistance in sinks is due to natural turnover and the clearance of resistant pathogens and low transmission rates. Thus, at least some part of the public health response should be to identify and focus control efforts on those institutions that are sources—usually places where antibiotics are heavily used (Hartley, Furuno et al. 2006). Important source institutions are hospitals, longterm care facilities, daycare centers, and prisons.

The source-sink dynamic interacts with underlying heterogeneity in the human population, and some subpopulations probably play a particularly important role in the spread of antibiotic resistance. One class of important players is health care workers, who are in the hospital every day (Trick, Weinstein et al. 2001). Since resistant bacteria can be extremely persistent, patients who are frequently hospitalized or who have very long hospital visits likely play an important role in the establishment of endemic populations (Furuno, McGregor et al. 2006).

In addition to health care workers and patients, the institution itself—the building, furniture, and equipment—can become contaminated with antibiotic-resistant bacteria, especially species that are able to persist for long periods in the environment. Some species of bacteria are very durable and can persist on bed frames, couches, medical instruments, and other objects, from which they can continue to contaminate the hands of health care workers and patients. Such contamination may play an important role in the establishment of endemic populations of resistance in hospitals and long-term care facilities (Hayden, Bonten et al. 2006).

Importantly, few antibiotics are used for just one bacterial pathogen, and few bacteria are resistant to just one antibiotic, but most studies of antibiotic resistance have focused on the “one-bug, one-drug” problem. Another conceptual model for the emergence of resistance is that antibiotic use has acted as a major perturbation to microbial communities, and that the emergence of resistance is the result of disturbing whole communities. In these disturbed environments, the bacteria that have thrived are those that are naturally invasive—the equivalent of weeds. This may account for the emergence of enterococci and VRE as important hospital-acquired infections in the 1980s (Box 2.7). Long-term changes in the frequency of Enterococcus faecalis and E. faecium and the increased frequency of Clostridium difficile infections may be important medical side effects of community-level changes that are driven by antibiotic use.

There are, of course, other ways that resistance can spread rapidly through a population. One is the accidental spread by hitchhiking on other successful genes. Since resistance genes are all likely to be successful in the same places, several antibiotic resistance genes can all travel together on the same mobile genetic element and contribute to the increasing frequency of bacteria that are resistant to more than one antibiotic. Another way is for resistance to be present in the founding member of a very successful clone; such founder effects may account for recent changes in the epidemiology of MRSA (Box 2.8). Since genes move among bacteria, some bacteria species that are not important pathogens are nevertheless clinically important because they act as a reservoir of resistance genes. The relative importance of all these mechanisms has not been fully explored.

The aim of control

When considering the control of antibiotic-resistant bacteria, what should be the goal? Given the complex biology and epidemiology of antibiotic resistance, there are several possible goals. One is to delay the emergence of antibiotic resistance. Another is to slow the spread with the long-term goal of reducing the frequency of resistance, or at least reducing the rate at which the frequency is increasing. A third goal is to reduce the number of infections with antibiotic-resistant bacteria. These goals often point to the same interventions for controlling resistant infections, but not always.

Reducing the frequency of resistance is not, in and of itself, the primary goal of policy. If all one wanted to do was reverse the spread of resistance, one could simply stop using all antibiotics. If no antibiotic resistance had emerged, newly evolved strains would certainly not have an advantage, so the frequency of antibiotic resistance would decline, albeit slowly, because of the biological cost of resistance. Antibiotics are valuable, however, because of their ability to treat infections. Although eliminating antibiotic use may be effective in minimizing resistance, this would be like never driving one’s car to avoid scratching it.

On the other hand, one could simply prevent as many infections as possible, but since some infection is inevitable, preventing infections is only a partially effective goal. If a policy does not reverse an increasing trend in the frequency of resistance, eventually all the infections will be resistant.

What is the right goal for a program to control antibiotic resistance? If we consider antibiotic effectiveness a scarce resource, the obvious goal is to maximize the total net benefit of an antibiotic in a population over time. The total benefit should be weighted toward treating people who are sick in the present and people with a critical need in the future, rather than people who might receive a marginal benefit in the present. The preference for discounting future resistance has a medical justification, in addition to the economic reasons for discounting the future. New therapies to treat infections may be available in the future that are not available today, so future infections are always discounted by the expectation that they will be treatable by other means.

The goal of maximizing the total net benefit of an antibiotic can be accomplished by delaying the emergence of resistance, by slowing the spread of resistance to reduce the frequency of antibiotic-resistant infections, and by reducing the total number of infections. To find the balance of strategies that would maximize the right objective function, some new analysis is required.

Controlling antibiotic use

The best and most obvious way to delay emergence and slow spread is to reduce selection by eliminating the use of antibiotics when they do not provide any medical benefit. Doctors often prescribe antibiotics unnecessarily for several reasons: 1) the patient and doctor have a different incentive to treat an infection than society; 2) there is always some uncertainty associated with diagnosing a medical problem; 3) reducing that uncertainty requires time and money; 4) all else equal, antibiotics are most likely to benefit a patient if the treatment is started early; and 5) patients want antibiotics when they are not needed, for various reasons (see Chapter 3 on patient and physician demand for antibiotics).

To put it another way, what happens when a patient presents with symptoms that are occasionally caused by bacteria but usually caused by something else? The doctor can either prescribe empiric therapy or wait until a diagnostic test confirms a bacterial infection. The decision to use an antibiotic immediately gives the patient something tangible to take away from the visit and makes him feel better, protects a doctor’s liability on the off-chance that the infection was bacterial, and eliminates the time, expense, and delay of an additional visit and laboratory diagnostic tests. On the other hand, diagnostic tests can help the doctor identify the cause of an infection and, especially in the case of inappropriate empiric therapy, choose more appropriate treatments. Thus, increased use of diagnostic tools can help reduce treatment failures from other causes and reduce mortality (Fagon, Chastre et al. 2000).

Offsetting those incentives and benefits are the cost of the antibiotic, the risk of an adverse reaction to the medicine, and the risk of resistance. All of these costs are mitigated for the doctor and patient: the doctor doesn’t pay for the antibiotic and the patient’s prescription drugs are often subsidized. Doctors are not typically regarded as liable for adverse reactions, and patients do not regard antibiotics as dangerous. Although the patient does increase his risk of antibiotic resistance in future infections from the increased resistance in his resident flora, that threat seems remote.

Cultural norms and other factors also influence the desire to take antibiotics, but the above caricature of the complex negotiation between doctor and patient illustrates several of the reasons why it is difficult to reduce unnecessary antibiotic use.

One suggestion is that delaying prescriptions for some kinds of infections could reduce antibiotic use without placing patients at elevated risk (Edwards, Dennison et al. 2003).

Another way to reduce the total selective pressure applied by today’s drugs is to shift away from broad-spectrum antibiotics to narrow-spectrum antibiotics. A broad-spectrum antibiotic is more likely to work against an infection with unidentified bacteria. Despite the advantages to the patient, the broadspectrum antibiotic selects for resistance in several bacteria species all at once. A narrow-spectrum antibiotic should limit the collateral damage. Unfortunately, there is not much research to support these assertions, since the effects of various antibiotics on the emergence of resistance within multiple species have not been studied.

Unnecessary antibiotic use and a shift from broad- to narrow-spectrum antibiotics would be facilitated if rapid diagnostic tests were available and affordable, and if the financial incentives to doctors and patients were changed to encourage their use. However, many bacteria that cause infections are not easily cultured and may not be detected with a rapid diagnostic test. Although some unidentifiable bacterial infections make up a small percentage of infections, they do occur. Rapid diagnostic tests could be improved with new technology, but the rapid tests are neither perfect nor inexpensive, and increasing testing can strain the microbiological resources of a hospital.

Finally, antibiotic resistance may be managed by manipulating the relative amounts of different antibiotics that are used. One proposed strategy is to rotate or cycle antibiotics. Although there are solid a priori arguments in favor of cycling antibiotics, a recent systematic review found insufficient evidence to evaluate cycling as a policy (Masterton 2005). Available ecological theory also suggests that concurrent use of multiple antibiotics would be more effective than cycling (Bonhoeffer, Lipsitch et al. 1997; Bergstrom, Lo et al. 2004), but there is very little high-quality empirical evidence to evaluate these strategies. A closely related strategy that reduces the number of antibiotic-resistant infections is to clear the colonizing bacteria, usually with some other antibiotic; for example, mupirocin is often used to eliminate colonization with MRSA in patients who are at risk of infection (Perl, Cullen et al. 2002). This is a useful strategy, but resistance to the alternative antibiotic is also a concern.

A controversial bit of advice that does not unambiguously stem the rising tide of antibiotics is to “take the full course of antibiotics.” Antibiotics should be taken if there is a danger that a partially treated infection will recrudesce, but what if a patient has started on antibiotics and a subsequent diagnostic test reveals that the infection is not bacterial? Should he continue to take a full course of antibiotics? At its extreme, the full course of antibiotics increases the total amount of antibiotics used and increases the possibilities for collateral damage. This advice clearly needs a more critical evaluation.

That raises the closely related issue of the design of antibiotic courses in general. Is the duration of a course of antibiotics longer than it needs to be to clear an infection? Once the risk of recrudescence has passed, so has the need for antibiotics, but continued use may contribute to the spread of resistance. Improving antibiotic dosing by taking short, intense courses could be an important part of limiting antibiotic use, but it would require substantial additional PK/PD research (for example, see Schrag et al. 2001). This suggests that antibiotics should be given more intensively and for shorter intervals, a general principle that is finding increasing application (Drusano 2003; Drusano 2005).

Controlling infection

Another strategy for controlling resistance is to reduce transmission of bacteria. Indeed, the Hospital Infection Control Practices Advisory Committee recommendations for controlling VRE include advice about reducing transmission of these pathogens from patient to patient (HICPAC 1995). Hospital infection control has at least two important goals, one of which is to protect individual patients from infections. This would presumably limit opportunities for the spread of VRE and eliminate outbreaks. A secondary effect may be to reduce the number of bacterial infections, and thus reduce the total amount of antibiotics used.

The other purpose of hospital infection control, one that is identified for the first time in this report, is to reduce the fitness of antibiotic-resistant bacteria. From the perspective of an individual human host, there are two ways to manipulate the among-host component of bacterial fitness: slow transmission or reduce persistence. An important principle is that antibiotic resistance can be controlled without eliminating all antibiotic use or completely isolating every patient. Antibiotic-resistant pathogens can be thought of in two ways, for the purposes of control: they can be considered a simple pathogen, or they can be considered competition for drug-sensitive bacteria. Either way, controlling the spread of resistance involves threshold phenomena: it is sufficient to reduce the rate of transmission such that every pathogen is spread from each person to less than one other person, or that the resistant strains spread less efficiently than sensitive strains.

That evolutionary perspective is critical. Any action taken to reduce transmission in some general way reduces transmission for drug-sensitive bacteria as well. Thus, to be most effective, hospital infection control should be applied selectively—that is, it should be applied more strenuously and effectively against resistant bacteria. In other words, the best hospital infection control policy is to be selectively clean.

What are the most effective ways to slow transmission? Since health care workers act as vectors that carry resistant bacteria among patients, via contaminated hands, clothing, or medical instruments, simple measures to reduce transmission include hand washing, gloves, gowns, and other barrier precautions. Other strategies include more careful decontamination of medical instruments and having health care workers trim their fingernails and remove neckties. Transmission can also be improved by limiting the number of contacts: isolating colonized or at-risk patients, reducing nurses’ workloads, or decreasing cohorting (i.e., reducing the number of patients seen by each nurse).

Another effective way of reducing resistance is to eliminate the resistant bacteria that have colonized the hospital environment. Some kinds of bacteria are extremely durable and persist on hospital equipment for weeks or months. The frequency of resistance can be reduced by improving cleaning procedures (Hayden, Bonten et al. 2006). Health care workers who may be carriers of resistance are yet another target (Lessing, Jordens et al. 1996; Lange, Morrissey et al. 2000; et al. 2001; Baran, Ramanathan et al. 2002).

Some hospital infection control measures have been difficult to implement or sustain, however. Improved hand washing is typically difficult to achieve unless hospitals make dramatic changes, such as altering hospital architecture to make hand-washing sinks readily available, or including hand washing as a performance measure linked to pay and promotion. Such practices would need to be enforced on doctors as well as nurses.

Another source of resistance is the hospital patients themselves. One of the major problems with hospital infection control is that it fails to identify many patients who are already colonized by antibiotic-resistant bacteria at the time of admission. A solution is to test patients on admission, a process known as active surveillance. It is, of course, expensive to test everyone and isolate patients, but studies suggest it would be cost-effective (Perencevich, Fisman et al. 2004). Patients can be isolated until they return a negative test or isolated after they test positive; the former strategy would be more effective but more expensive. A simple way to reduce the number of patients to be tested while still identifying most carriers is to focus on patients who have significant risk factors, such as having been recently hospitalized (Furuno, McGregor et al. 2006), admitted from a long-term care facility, or prescribed antibiotics.

Given the costs of active surveillance, an alternative strategy would be for hospitals to share information with each other about the antibiotic-resistant bacteria that are known to colonize a patient. The development of a personal electronic medical history that includes laboratory test results, previous antibiotics used, and other information would make this strategy efficient and offer other benefits, such as providing doctors with a more complete patient history. The record could be useful for hospitals to identify and isolate potential carriers.

Fewer options exist for reducing the persistence time of resistant bacteria, except possibly through selective decontamination. Some other interventions may be found through focused research on the ecological relationships among bacteria, which could reduce the prevalence of colonization with medically important species.

Novel ecological strategies

Ecological strategies might also counter the effects of antibiotic use. If antibiotic resistance has spread because antibiotic use has opened a niche, that niche could be filled with something else. In general, this something else would have to be another bacterial species, one that is not pathogenic. Harmless bacteria that are taken for such purposes are called probiotics. In fact, probiotics are a common part of informal health care: people often eat live-culture yogurt after taking antibiotics to restore their normal flora. Probiotics have a place in treating bacterial overgrowth, but the therapeutic and public health value of probiotics to reduce antibiotic resistance is not well studied, and early results are not promising (Lund, Edlund et al. 2000; Lund, Adamsson et al. 2002).

Another possibility is to exploit biological control agents. For example, a bacteriophage that attacks particular strains of bacteria could be engineered and spread around to decontaminate a hospital or to reduce bacterial colonization of the skin. These strategies are largely speculative.

A successful policy will have a solid epidemiological basis but must also follow economic principles. Since antibiotic resistance spreads among hospitals, states, and nations, any program used by one agent to control the spread of resistance within its own borders can be undermined by the inaction of a neighbor. Players acting in their own selfinterest, moreover, may not behave in a way that benefits the common good. Without coordination, some players may choose to free-ride on the investments of their neighbors. For example, hospitals that share patients may choose to free-ride on the infection-control policies of other nearby hospitals (Smith, Levin et al. 2005).When all players adopt that strategy, the result is a Nash equilibrium—a perfectly rational but unfortunate global strategy.

Success stories

The policies for controlling the spread of resistance in the United States and across the world have largely failed, but a few success stories do exist. Two are the Dutch experience with MRSA control, and VRE control in the Siouxland (also discussed in Chapter 4). Both of these approaches were very aggressive, and both involved coordination at some large regional level.

In the Netherlands, patients admitted to hospitals from outside the country were automatically quarantined. When transmission did occur, radical measures were taken to control transmission; when transmission was documented, hospital units were shut down. Dutch citizens were not tracked individually, but they were known to be safe because the national policy was universally followed and MRSA was so rare. A few expensive outbreaks of MRSA were identified with Dutch citizens who became carriers while hospitalized outside the Dutch system (Verhoef, Beaujean et al. 1999). Despite the inconvenience and the huge expense of the “search-and-destroy” strategy, it was cost-effective because it reduced the number of infections of both VRE and MRSA (Vriens, Blok et al. 2002).

One notable effect of the Dutch strategy was to reduce transmission of S. aureus and select against the most transmissible clone until it was eliminated from the Netherlands (Verhoef, Beaujean et al. 1999). Thus, as time went on, hospital infection control measures had the added benefit of selecting for bacterial pathogens that were easier to control.

In Siouxland—where Iowa, Nebraska, and South Dakota meet—several hospitals coordinated their efforts. They tracked patients who had become carriers and quarantined them when they entered other hospitals. This strategy, combined with patient isolation measures, led to decreases in the transmission and prevalence of VRE (Ostrowsky, Trick et al. 2001; Sohn, Ostrowsky et al. 2001).

Major shifts in the epidemiology of disease have been caused by the evolution of pathogens unrelated to resistance, and these should be regarded as important side effects—both positive and negative—of the use of antibiotics and of enlightened infection control. As mentioned, the radical hospital infection control procedures in the Netherlands eliminated one of the most transmissible types of S. aureus. And acute rheumatic fever is now rare in the United States, mostly because of treatment of Group A streptococcal infections with penicillin (Krause 2002). An increasing incidence of Clostridium difficile infections, however, may be a negative side effect of antibiotic use.




The spread of resistance is an ecological problem involving competition between sensitive and resistant bacteria. Resistant bacteria may have inherent disadvantages or fitness costs when they first evolve, but these disadvantages may evolve away over time. Antibiotic-resistant bacteria are infectious agents, and the conditions for spread involve thresholds. Because of these thresholds and the complex ecology of resistant bacteria, the design and control of antibiotic resistance are difficult to study, and interpretation of data is often not straightforward (Lipsitch, Bergstrom et al. 2000). The success of various control measures may not become apparent unless they reduce selection below a threshold, and the effects may not become apparent until the ecological reservoir of resistance is reduced—a process that could take years.

There are real obstacles to changing the way antibiotics are used, and real difficulties in slowing the spread of resistance. Coordination and investment have been insufficient. Many potential solutions to the problem of resistance exist, but the incentives of the agents are not aligned, and it is unlikely that any of these solutions will work without structural changes to the way all the agents deal with antibiotic resistance. Cooperation among institutions is difficult, and active surveillance to identify carriers who are entering an institution is expensive for single hospitals. Despite all this, there have been some successes when efforts were coordinated across wide regions. Electronic records, combined with screening algorithms on admission to recognize potential carriers, would improve coordination and have other benefits as well. Most of these problems and potential solutions are dealt with
in greater detail in other chapters.

The real costs and benefits of various control measures have not been properly quantified in a controlled environment. Further study and concerted efforts to implement policy measures at scale are necessary to deal with the problem of antibiotic resistance.




Agvald-Ohman, C., B. Lund, et al. (2004). “Multiresistant Coagulase- Negative Staphylococci Disseminate Frequently Between Intubated Patients in a Multidisciplinary Intensive Care Unit.” Critical Care 8(1): 42-47.

Aiello, A. E., F. D. Lowy, et al. (2006). “Meticillin-Resistant Staphylococcus aureus Among US Prisoners and Military Personnel: Review and Recommendations for Future Studies.” Lancet Infectious Diseases 6(6): 335-341.

Andersson, D. I. (2006). “The Biological Cost of Mutational Antibiotic Resistance: Any Practical Conclusions?” Current Opinion Microbiology 9(5): 461-5.

Andersson, D. I. and B. R. Levin. (1999). “The Biological Cost of
Antibiotic Resistance.” Current Opinion in Microbiology 2(5): 489-493.

Antia, R., R. R. Regoes, et al. (2003). “The Role of Evolution in the Emergence of Infectious Diseases.” Nature 426(6967): 658-661.

Aubry-Damon, H., K. Grenet, et al. (2004). “Antimicrobial Resistance in Commensal Flora of Pig Farmers.” Emerging Infectious Diseases 10(5): 873-879.

Austin, D. J., M. Kakehashi, et al. (1997). “The Transmission Dynamics of Antibiotic-Resistant Bacteria: The Relationship Between Resistance in Commensal Organisms and Antibiotic Consumption.” Proceedings of the Royal Society of London. Series B 264: 1629-1638.

Austin, D. J., K. G. Kristinsson, et al. (1999). “The Relationship Between the Volume of Antimicrobial Consumption in Human Communities and the Frequency of Resistance.” Proceedings of the National Academy of Sciences of the United States of America 96: 1152-1156.

Baran, J., J. Ramanathan, et al. (2002). ”Stool Colonization with Vancomycin-Resistant Enterococci in Health care Workers and their Households.” Infection Control and Hospital Epidemiology 23: 23-26.

Becker, N. (1989). Analysis of Infectious Disease Data. London: Chapman and Hall.

Bergstrom, C. T., M. Lo, et al. (2004). “Ecological Theory Suggests that Antimicrobial Cycling Will Not Reduce Antimicrobial Resistance in Hospitals.” Proceedings of the National Academy of Sciences of the United States of America 101: 13285-13290.

Bjorkman, J. and D. I. Andersson. (2000). “The Cost of Antibiotic Resistance from a Bacterial Perspective.” Drug Resistance Updates 3(4): 237-245.

Bonhoeffer, S., M. Lipsitch, et al. (1997). “Evaluating Treatment Protocols to Prevent Antibiotic Resistance.” Proceedings of the National Academy of Sciences of the United States of America 94: 12106-12111.

Bonten, M. J., M. K. Hayden, et al. (1996). ”Epidemiology of Colonisation of Patients and Environment with Vancomycin-Resistant Enterococci.” Lancet 348: 1615-1619.

Bonten, M. J., S. Slaughter, et al. (1998). “The Role of Colonization Pressure in the Spread of Vancomycin-Resistant Enterococci: An Important Infection Control Variable.” Archives of Internal Medicine 158: 1127-1132.

Bonten, M. J., R. Willems, et al. (2001). “Vancomycin-Resistant
Enterococci: Why Are They Here, and Where Do They Come From?” The Lancet Infectious Diseases 1: 314-325.

Carbon, C. (1999). “Costs of Treating Infections Caused by Methicillin-Resistant Staphylococci and Vancomycin-Resistant Enterococci.” The Journal of Antimicrobial Chemotherapy 44(Suppl A): 31-36.

Carmeli, Y., G. M. Eliopoulos, et al. (2002). ”Antecedent Treatment with Different Antibiotic Agents as a Risk Factor for Vancomycin-Resistant Enterococcus.” Emerging Infectious Diseases 8(8): 802-7.

CDC. (2002a). “Staphylococcus aureus Resistant to Vancomycin—United States, 2002.” MMWR. Morbidity and Mortality Weekly Report 51: 565-567. Centers for Disease Control and Prevention.

———. (2002b). “Public Health Dispatch: Vancomycin-Resistant Staphylococcus aureus—Pennsylvania, 2002.” MMWR. Morbidity and Mortality Weekly Report 51: 902.

———. (2003). “Methicillin-Resistant Staphylococcus aureus Infections in Correctional Facilities—Georgia, California, and Texas, 2001-2003.” MMWR. Morbidity and Mortality Weekly Report 52(41): 992-996.

———. (2004). “National Nosocomial Infections Surveillance (NNIS) System Report, Data Summary from January 1992 Through June 2004, Issued October 2004.” American Journal of Infection Control 32: 470-485.

Cooper, B. S., G. F. Medley, et al. (1999). “Preliminary Analysis of the Transmission Dynamics of Nosocomial Infections: Stochastic and Management Effects.” The Journal of Hospital Infection 43: 131-147.

Cooper, B. S., G. F. Medley, et al. (2004). “Methicillin-Resistant Staphylococcus aureus in Hospitals and the Community: Stealth Dynamics and Control Catastrophes.” Proceedings of the National Academy of Sciences of the United States of America 101: 10223-10228.

Costerton, J. W., P. S. Stewart, et al. (1999). “Bacterial Biofilms: A Common Cause of Persistent Infections.” Science 284(5418): 1318-1322.

Courvalin, P. (2000). “Will Avilamycin Convert Ziracine into Zerocine?” Emerging Infectious Diseases 6(5): 558.

Dall’Antonia, M., P. G. Coen, et al. (2005). “Competition Between Methicillin-Sensitive and -Resistant Staphylococcus aureus in the Anterior Nares.” The Journal of Hospital Infection 61(1): 62-7.

Davies, J. (1994). “Inactivation of Antibiotics and the Dissemination of Resistance Genes.” Science 264: 375-382.

Donskey, C. J., T. K. Chowdhry, et al. (2000). “Effect of Antibiotic Therapy on the Density of Vancomycin-Resistant Enterococci in the Stool of Colonized Patients.” The New England Journal of Medicine 28: 1925-1932.

Donskey, C. J., J. A. Hanrahan, et al. (1999). “Effect of Parenteral Antibiotic Administration on Persistence of Vancomycin-Resistant Enterococcus faecium in the Mouse Gastrointestinal Tract.” The Journal of Infectious Diseases 180: 384-390.

———. (2000). “Effect of Parenteral Antibiotic Administration on the Establishment of Colonization with Vancomycin-Resistant Enterococcus faecium in the Mouse Gastrointestinal Tract.” The Journal of Infectious Diseases 181: 1830-1833.

Dowell, S. F. and B. Schwartz. (1997). “Resistant Pneumococci:
Protecting Patients Through Judicious Use of Antibiotics.” American Family Physician 55(5): 1647-54, 1657-8.

Drusano, G. L. (2003). “Prevention of Resistance: A Goal for Dose Selection for Antimicrobial Agents.” Clinical Infectious Diseases 36(Suppl 1): 42-50.

———. (2005). “Infection Site Concentrations: Their Therapeutic
Importance and the Macrolide and Macrolide-Like Class of Antibiotics.” Pharmacotherapy 25(12 Pt 2): 150S-158S.

Dykes, G. A. (1995). “Bacteriocins: Ecological and Evolutionary
Significance.” Trends in Ecology & Evolution 10(5): 186-189. Dykes, G. A. and J. W. Hastings. (1997). “Selection and Fitness in Bacteriocin-Producing Bacteria.” Proceedings of the Royal Society of London. Series B 264(1382): 683-687.

Edwards, M., J. Dennison, et al. (2003). “Patients’ Responses to Delayed Antibiotic Prescription for Acute Upper Respiratory Tract Infections.” The British Journal of General Practice 53(496): 845-850.

Elizaga, M. L., R. A. Weinstein, et al. (2002). “Patients in Long-Term Care Facilities: A Reservoir for Vancomycin-Resistant Enterococci.” Clinical Infectious Diseases 34: 441-446.

Fagon, J. Y., J. Chastre, et al. (2000). “Invasive and Noninvasive Strategies for Management of Suspected Ventilator-Associated Pneumonia. A Randomized Trial.” Annals of Internal Medicine 132(8): 621-30.

Furuno, J. P., J. C. McGregor, et al. (2006). “Identifying Groups at High Risk for Carriage of Antibiotic-Resistant Bacteria.” Archives of Internal Medicine 166(5): 580-585.

Fux, C. A., P. Stoodley, et al. (2003). “Bacterial Biofilms: A Diagnostic and Therapeutic Challenge.” Expert Review of Anti-Infective Therapy 1(4): 667-683.

Hartley, D. M., J. P. Furuno, et al. (2006). “The Role of Institutional Epidemiologic Weight in Guiding Infection Surveillance and Control in Community and Hospital Populations.” Infection Control and Hospital Epidemiology 27(2): 170-174.

Hayden, M. K., M. J. M. Bonten, et al. (2006). “Reduction in Acquisition of Vancomycin-Resistant Enterococcus After Enforcement of Routine Environmental Cleaning Measures.” Clinical Infectious Diseases 42(11): 1552-1560.

Henning, K. J., H. Delencastre, et al. (1996). “Vancomycin-Resistant Enterococcus faecium on a Pediatric Oncology Ward: Duration of Stool Shedding and Incidence of Clinical Infection.” The Pediatric Infectious Disease Journal 15: 845-847.

Herold, B. C., L. C. Immergluck, et al. (1998). “Community-Acquired Methicillin-Resistant Staphylococcus aureus in Children with No Identified Predisposing Risk.” JAMA 279(8): 593-598.

HICPAC. (1995). “Recommendations for Preventing the Spread of Vancomycin Resistance. Recommendations of the Hospital Infection Control Practices Advisory Committee (HICPAC).” MMWR. Recommendations and Reports 44(RR-12): 13-Jan. Hospital Infection Control Practices Advisory Committee.

Jack, R. W., J. R. Tagg, et al. (1995). ”Bacteriocins of Gram-Positive Bacteria.” Microbiological Reviews 59(2): 171-200.

Kluytmans, J., A. v. Belkam, et al. (1997). “Nasal Carriage of Staphylococcus aureus: Epidemiology, Underlying Mechanisms, and Associated Risks.” Clinical Microbiology Reviews 10: 505-520.

Kluytmans-Vandenbergh, M. F. Q. and J. A. J. W. Kluytmans. (2006). “Community-Acquired Methicillin-Resistant Staphylococcus aureus: Current Perspectives.” Clinical Microbiology and Infection 12 Suppl 1: 15-Sep.

Krause, R. M. (2002). “A Half-Century of Streptococcal Research: Then & Now.” The Indian Journal of Medical Research 115: 215-241.

Kummerer, K. (2004). “Resistance in the Environment.” The Journal of Antimicrobial Chemotherapy 54(2): 311-20.

Lange, C. G., A. B. Morrissey, et al. (2000). “Point-Prevalence of
Contamination of Health care Workers’ Stethoscopes with Vancomycin-Resistant Enterococci at Two Teaching Hospitals in Cleveland, Ohio.” Infection Control and Hospital Epidemiology 21: 756.

Lessing, M. P., J. Z. Jordens, et al. (1996). ”When Should Health care Workers Be Screened for Methicillin-Resistant Staphylococcus aureus?” The Journal of Hospital Infection 34: 205-210.

Levin, B. R. (2001). “Minimizing Potential Resistance: A Population Dynamics View.” Clinical Infectious Diseases 33(Suppl 3): S161-S169.

———. (2004). “Microbiology. Noninherited Resistance to Antibiotics.” Science 305(5690): 1578-1579.

Levin, B. R. and R. Antia. (2001). “Why We Don’t Get Sick: The Within-Host Population Dynamics of Bacterial Infections.” Science 292(5519): 1112-1115.

Levin, B. R., V. E. Perrot, et al. (2000). ”Compensatory Mutations, Antibiotic Resistance and the Population Genetics of Adaptive Evolution in Bacteria.” Genetics 154: 985-997.

Lipsitch, M. (2001). “Measuring and Interpreting Associations Between Antibiotic Use and Penicillin Resistance in Streptococcus pneumoniae.” Clinical Infectious Diseases 32: 1044-1054.

———. (2001). “The Rise and Fall of Antimicrobial Resistance.” Trends in Microbiology 9: 438-444.

Lipsitch, M., C. T. Bergstrom, et al. (2000). ”The Epidemiology of Antibiotic Resistance in Hospitals: Paradoxes and Prescriptions.” Proceedings of the National Academy of Sciences of the United States of America 97: 1938-1943.

Lipsitch, M. and M. H. Samore. (2002). “Antimicrobial Use and
Antimicrobial Resistance: A Population Perspective.” Emerging Infectious Diseases 8: 347-354.

Lund, B., I. Adamsson, et al. (2002). “Gastrointestinal Transit Survival of an Enterococcus faecium Probiotic Strain Administered With or Without Vancomycin.” International Journal of Food Microbiology 77(2-Jan): 109-115.

Lund, B., C. Agvald-Ohman, et al. (2002). ”Frequent Transmission of Enterococcal Strains Between Mechanically Ventilated Patients Treated at an Intensive Care Unit.” Journal of Clinical Microbiology 40(6): 2084-2088.

Lund, B., C. Edlund, et al. (2000). “Impact on Human Intestinal Microflora of an Enterococcus faecium Probiotic and Vancomycin.” Scandinavian Journal of Infectious Diseases 32(6): 627-632.

Masterton, R. G. (2005). “Antibiotic Cycling: More Than It Might Seem?” The Journal of Antimicrobial Chemotherapy 55(1): 5-Jan.

McGowan, J. E., Jr. (1983). “Antimicrobial Resistance in Hospital Organisms and Its Relation to Antibiotic Use.” Reviews of Infectious Diseases 5(6): 1033-48.

Monnet, D. L. and J. M. Lopez-Lozano. (2005). “Relationship Between Antibiotic Consumption and Resistance in European Hospitals.” Medecine et Maladies Infectieuses 35 Suppl 2: S127-8.

Nicolle, L. E. (2001). “Preventing Infections in Non-Hospital Settings: Long-Term Care.” Emerging Infectious Diseases 7(2): 205-207.

O’Brien, T. F. (2002). “Emergence, Spread, and Environmental Effect of Antimicrobial Resistance: How Use of an Antimicrobial Anywhere Can Increase Resistance to Any Antimicrobial Anywhere Else.” Clinical Infectious Diseases. 34(Suppl 3): S78-S84.

Okeke, I. N. and R. Edelman. (2001). “Dissemination of Antibiotic-Resistant Bacteria Across Geographic Borders.” Clinical Infectious Diseases. 33: 364-9.

Ostrowsky, B. E., W. E. Trick, et al. (2001). “Control of Vancomycin-Resistant Enterococcus in Health Care Facilities in a Region.” The New England Journal of Medicine 344: 1427-33.

Paterson, D. L. and L. B. Rice. (2003). “Empirical Antibiotic Choice for the Seriously Ill Patient: Are Minimization of Selection of Resistant Organisms and Maximization of Individual Outcome Mutually Exclusive?” Clinical Infectious Diseases 36(8): 1006-1012.

Pelupessy, I., M. J. Bonten, et al. (2002). “How to Assess the Relative Importance of Different Colonization Routes of Pathogens Within Hospital Settings.” Proceedings of the National Academy of Sciences of the United States of America 99: 5601-5605.

Perencevich, E. N., D. N. Fisman, et al. (2004). “Projected Benefits of Active Surveillance for Vancomycin-Resistant Enterococci in Intensive Care Units.” Clinical Infectious Diseases 38(8): 1108-15.

Perl, T. M., J. J. Cullen, et al. (2002). ”Intranasal Mupirocin to Prevent Postoperative Staphylococcus aureus Infections.” The New England Journal of Medicine 346: 1871-1877.

Phillips, I., M. Casewell, et al. (2004). “Does the Use of Antibiotics in Food Animals Pose a Risk to Human Health? A Critical Review of Published Data.” The Journal of Antimicrobial Chemotherapy 53: 28-52.

Rice, L. B. (2001). “Emergence of Vancomycin-Resistant Enterococci.” Emerging Infectious Diseases 7: 183-187.

Riley, M. A. and J. E. Wertz. (2002). “Bacteriocins: Evolution, Ecology, and Application.” Annual Review of Microbiology 56: 117-137.

Rotun, S. S., V. McMath, et al. (1999). “Staphylococcus aureus with Reduced Susceptibility to Vancomycin Isolated From a Patient with Fatal Bacteremia.” Emerging Infectious Diseases 5(1): 147-149.

Rubin, R. J., C. A. Harrington, et al. (1999). “The Economic Impact of Staphylococcus aureus in New York City Hospitals.” Emerging Infectious Diseases 5(1): 9-17.

Saravolatz, L. D., N. Markowitz, et al. (1982). “Methicillin-Resistant Staphylococcus aureus. Epidemiologic Observations During a Community-Acquired Outbreak.” Annals of Internal Medicine 96(1): 16-Nov.

Saravolatz, L. D., D. J. Pohlod, et al. (1982). “Community-Acquired Methicillin-Resistant Staphylococcus aureus Infections: A New Source for Nosocomial Outbreaks.” Annals of Internal Medicine 97(3): 325-329.

Schrag, S.J., C. Pena, et al. (2001). “Effect of Short-Course, High-Dose Amoxicillin Therapy on Resistant Pneumococcal Carriage: A Randomized Trial.” JAMA 286(1): 49-56.

Smith, D. L., J. Dushoff, et al. (2004). “Persistent Colonization and the Spread of Antibiotic Resistance in Nosocomial Pathogens: Resistance Is a Regional Problem.” Proceedings of the National Academy of Sciences of the United States of America 101: 3709-3714.

Smith, D. L., J. Dushoff, et al. (2005). “Agricultural Antibiotics and Human Health.” PLoS Medicine 2(8): e232.

Smith, D. L., A. D. Harris, et al. (2002). “Animal Antibiotic Use Has an Early But Important Impact on the Emergence of Antibiotic Resistance in Human Commensal Bacteria.” Proceedings of the National Academy of Sciences of the United States of America 99(9): 6434-9.

Smith, D. L., S. A. Levin, et al. (2005). “Strategic Interactions in Multi-Institutional Epidemics of Antibiotic Resistance.” Proceedings of the National Academy of Sciences of the United States of America 102: 3153-3158.

Sohn, A. H., B. E. Ostrowsky, et al. (2001). “Evaluation of a Successful Vancomycin-Resistant Enterococcus Prevention Intervention in a Community of Health Care Facilities.” American Journal of Infection Control 29: 53-7.

Trick, W. E., M. J. Kuehnert, et al. (1999). "Regional Dissemination of Vancomycin-Resistant Enterococci Resulting from Interfacility Transfer of Colonized Patients." The Journal of Infectious Diseases 180: 391-396.

Trick, W. E., R. A. Weinstein, et al. (2001). “Colonization of Skilled-Care Facility Residents with Antimicrobial-Resistant Pathogens.” Journal of the American Geriatrics Society 49: 270-276.

Verhoef, J., D. Beaujean, et al. (1999). “A Dutch Approach to Methicillin-Resistant Staphylococcus aureus.” European Journal of Clinical Microbiology & Infectious Diseases 18: 461-466.

Vriens, M., H. Blok, et al. (2002). ”Costs Associated With a Strict Policy to Eradicate Methicillin-Resistant Staphylococcus aureus in a Dutch University Medical Center: A 10-year Survey.” European Journal of Clinical Microbiology & Infectious Diseases 21: 782-786.

Weinstein, R. A. (2001). “Controlling Antimicrobial Resistance in Hospitals: Infection Control and Use of Antibiotics.” Emerging Infectious Diseases 7: 188-192.

Witte, W. (1998). “Medical Consequences of Antibiotic Use in Agriculture.” Science 279: 996-997.

Wiuff, C., R. M. Zappala, et al. (2005). “Phenotypic Tolerance: Antibiotic Enrichment of Noninherited Resistance in Bacterial Populations.” Antimicrobial Agents and Chemotherapy 49(4): 1483-1494.




1 For bacteria to reach other hosts, they must be shed from carriers. Shedding means that bacteria are broadcast into the environment surrounding a host.

2 Broadly speaking, pathogens are classified as bacteria, eukaryotes, or viruses. Eukaryotic pathogens, including fungi, various types of intestinal worms, and Plasmodium falciparum (the parasite that causes malaria), have a nucleus, they are diploid, and they reproduce sexually. Viruses are bits of genetic material encapsulated in a protein coat, and by most criteria, they are not considered to be alive. Some viruses are composed of several pieces, so it is possible for the genetic material to recombine; influenza is a notable example. Compared with most eukaryotes and viruses, bacteria are extremely promiscuous—they have sex in more ways and with a broader diversity of organisms.