Chemotherapy of Microbial Diseases

Antimicrobial agents, the general classes of these drugs, and their mechanisms of action and mechanisms of bacterial resistance are reviewed in this chapter. The principles that are important for the selection of the appropriate antibiotic, the use of antibiotic combinations, and the role of chemoprophylaxis are discussed. This chapter presents both a philosophical and a practical approach to the appropriate use of antimicrobial agents as well as a discussion of the factors that influence the outcome of such treatment. Also emphasized is the frequent misuse of antimicrobial agents due to lack of identification of the responsible microorganism, leading in some cases to superinfection.

Optimal and judicious selection of antimicrobial agents for the therapy of infectious diseases requires clinical judgment and detailed knowledge of pharmacological and microbiological factors. Unfortunately, the decision to use antibiotics frequently is made lightly, without regard to the potential infecting microorganism or to the pharmacological features of the drug. Antibiotics are used in three general waysas empirical therapy, as definitive therapy, and as prophylactic or preventive therapy. When used as empirical, or initial, therapy, the antibiotic must 'cover' all of the likely pathogens, since the infecting organism(s) has not yet been defined. Combination therapy or treatment with a single broad-spectrum agent often is employed. However, once the infecting microorganism is identified, definitive antimicrobial therapy should be instituteda narrow-spectrum, low-toxicity regimen to complete the course of treatment. When an antimicrobial agent is indicated, the goal is to choose a drug that is selectively active for the most likely infecting microorganism(s) and that has the least potential to cause toxicity or allergic reactions in the individual being treated (see Table 431).

The first decision to be made is whether or not administration of an antimicrobial agent is even indicated. The reflex action of many physicians is to associate fever with treatable infections and prescribe antimicrobial therapy without further evaluation. This practice is irrational and potentially dangerous. The diagnosis may be masked if appropriate cultures are not obtained prior to therapy. Antibiotics can cause serious toxicity, and injudicious use of antimicrobial agents promotes selection of resistant microorganisms. Of course, definitive identification of a bacterial infection before treatment is initiated often is not possible. In the absence of a clear indication, antibiotics often may be used if disease is severe and if it seems likely that withholding therapy will result in failure to manage a potentially life-threatening infection.

Initiation of optimal empirical antibiotic therapy requires a knowledge of the most likely infecting microorganisms and their susceptibilities to antimicrobial drugs. A number of techniques are helpful in the selection of an antibiotic regimen. The clinical picture may suggest the specific microorganism. Knowledge of the microorganisms most likely to cause specific infections in a given host is essential. In addition, simple and rapid laboratory techniques are available for the examination of infected tissues. The most valuable and time-tested method for immediate identification of bacteria is the examination of the infected secretion or body fluid with Gram's stain. Tests such as this one help to narrow the list of potential pathogens and permit more rational selection of initial antibiotic therapy. However, in most situations, identification of the morphology of the infecting organism is not adequate to arrive at a specific bacteriological diagnosis, and the selection of a single narrow-spectrum antibiotic may be inappropriate, particularly if the infection is life-threatening. Broad antimicrobial coverage is then indicated, pending isolation and identification of the microorganism. Whenever the clinician is faced with initiating therapy on a presumptive bacteriological diagnosis, cultures of the presumed site of infection and blood, if bacteremia is a possibility, should be taken prior to the institution of drug therapy. For definitive therapy, the regimen should be changed to a more specific (narrow-spectrum) antimicrobial agent once an organism has been isolated and results of susceptibility tests are known.

Testing for Microbial Sensitivity to Antimicrobial Agents

There may be wide variations in the susceptibility of different strains of the same bacterial species to antibiotics. Information about the pattern of sensitivity of the infecting microorganism is important for appropriate drug selection. Several tests are now available for determination of bacterial sensitivity to antimicrobial agents. The most commonly used are disk-diffusion, agar- or broth-dilution tests, and automated test systems.

The disk-diffusion technique provides only qualitative or semiquantitative information on the susceptibility of a given microorganism to a given antibiotic. The test is performed by applying commercially available filter-paper disks impregnated with a specific amount of the drug onto an agar surface, over which a culture of the microorganism has been streaked. After 18 to 24 hours of incubation, the size of a clear zone of inhibition around the disk is measured. The diameter of the zone depends upon the activity of the drug against the test strain. Standardized values for zone sizes for each bacterial species and each antibiotic permit classification of the clinical isolate as resistant, intermediate, or susceptible.

Dilution tests employ antibiotics in serially diluted concentrations in solid agar or broth media containing a culture of the test microorganism. The lowest concentration of the agent that prevents visible growth after 18 to 24 hours of incubation is known as the minimal inhibitory concentration (MIC), and the lowest concentration that results in a 99.9% decline in bacterial numbers is known as the minimal bactericidal concentration (MBC). The value of the MBC as a clinical test has not been established.

Automated systems also use a broth-dilution method. The optical density of a broth culture of the clinical isolate incubated in the presence of drug is measured by absorbance densitometry. If the density of the culture exceeds a threshold optical density, then growth has occurred at that concentration of drug. The MIC is the concentration at which the optical density remains below the threshold.

Pharmacokinetic Factors

In vitro activity, although critical, is only a guide as to whether or not an antibiotic is likely to be effective in an infection. Successful therapy also depends upon achieving a drug concentration that is sufficient to inhibit or kill bacteria at the site of the infection without harming the host. To accomplish this therapeutic goal, several pharmacokinetic and host factors must be evaluated.

The location of the infection may, to a large extent, dictate the choice of drug and the route of administration. The minimal drug concentration achieved at the infected site should be approximately equal to the MIC for the infecting organism, although in most instances it is advisable to achieve multiples of this concentration if possible. However, there is evidence to suggest that even subinhibitory concentrations of antibiotics may enhance phagocytosis (Nosanchuk et al., 1999) and may be effective. Although these and related observations may explain why some infections are cured even when inhibitory concentrations are not achieved, it should be the aim of antimicrobial therapy to produce antibacterial concentrations of drug at the site of infection during the dosing interval. This can be achieved only if the pharmacokinetic and pharmacodynamic principles presented in Chapters 1: Pharmacokinetics: The Dynamics of Drug Absorption, Distribution, and Elimination and 2: Pharmacodynamics: Mechanisms of Drug Action and the Relationship Between Drug Concentration and Effect are understood and employed.

Access of antibiotics to sites of infection depends on multiple factors. If the infection is in the cerebrospinal fluid (CSF), the drug must pass the blood-brain barrier, and many antimicrobial agents that are polar at physiological pH do so poorly; some, such as penicillin G, are actively transported out of the CSF by an anion transport mechanism in the choroid plexus. The concentrations of penicillins and cephalosporins in the CSF are usually only 0.5% to 5% of steady-state concentrations determined simultaneously in plasma. However, the integrity of the blood-brain barrier is diminished during active bacterial infection; tight junctions in cerebral capillaries open, leading to a marked increase in the penetration of even polar drugs (Quagliarello and Scheld, 1997). As the infection is eradicated and the inflammatory reaction subsides, penetration reverts toward normal. Since this may occur while viable microorganisms persist in the CSF, drug dosage should not be reduced as the patient improves until the CSF is presumed or proven to be sterile.

Penetration of drugs into infected loci almost always depends on passive diffusion. The rate of penetration is thus proportional to the concentration of free drug in the plasma or extracellular fluid. Drugs that are extensively bound to protein thus may not penetrate to the same extent as those that are bound to a lesser extent. Drugs that are highly protein bound also may have reduced activity, because only the unbound fraction of drug is free to interact with its target.

Traditionally, the dose and dosing frequency of antibiotics have been selected to achieve antibacterial activity at the site of infection for most of the dosing interval. However, controversy exists as to whether the therapeutic effect achieved from relatively constant antibacterial activity is superior to that from high peak concentrations followed by periods of subinhibitory activity. To a certain extent, this depends upon whether a drug exhibits concentration-dependent or time-dependent growth inhibition (Craig, 1998). The activity of -lactam antibiotics, for example, is primarily time-dependent, whereas that of aminoglycosides is concentration-dependent. Activity also may depend upon the specific organism and the site of infection. Studies in animals with meningitis suggest that pulse dosing (intermittent administration) of -lactam antibiotics may be more efficient (equivalent efficacy from less drug) (Tuber et al., 1989); it appears that constant activity is superior in other experimental infections. Experimental data suggest that aminoglycosides are at least as efficacious, and are less toxic, when given in a single, large daily dose as when given more frequently (Gilbert, 1991; Barclay et al., 1999). Studies in patients also suggest that continuous administration of aminoglycosides may cause unnecessary toxicity.

Knowledge of the status of the individual patient's mechanisms for elimination of drugs also is essential, especially when excessive plasma or tissue concentrations of the drugs may cause serious toxicity. Most antimicrobial agents and their metabolites are eliminated primarily by the kidneys. Specific nomograms are available to facilitate adjustment of dosage of many such agents in patients with renal insufficiency. These are discussed in the chapters dealing with the individual drugs and in Appendix II. One must be particularly careful when using aminoglycosides, vancomycin, or flucytosine in patients with impaired renal function, since these drugs are eliminated exclusively by renal mechanisms, and their toxicity appears to correlate with their concentration in plasma and tissue. If renal toxicity of a drug that is cleared by the kidney occurs and care is not exercised, a vicious cycle may ensue. For drugs that are metabolized or excreted by the liver (erythromycin, chloramphenicol, metronidazole, clindamycin), dosages may have to be reduced in patients with hepatic failure.

Route of Administration

The discussion of choice of routes of administration that appears in Chapter 1: Pharmacokinetics: The Dynamics of Drug Absorption, Distribution, and Elimination of course applies to antimicrobial agents. While oral administration is preferred whenever possible, parenteral administration of antibiotics usually is recommended in seriously ill patients in whom predictable concentrations of drug must be achieved. Specific factors that govern the choice of route of administration for individual agents are discussed in the chapters that follow.

Host Factors

Innate host factors can be the prime determinants not only of the type of drug selected but also of its dosage, route of administration, risk and nature of untoward effects, and therapeutic effectiveness.

Host Defense Mechanisms

A critical determinant of the therapeutic effectiveness of antimicrobial agents is the functional state of host defense mechanisms. Both humoral and cellular immunity are important. Inadequacy of type, quality, and quantity of the immunoglobulins; alteration of the cellular immune system; or a qualitative or quantitative defect in phagocytic cells may result in therapeutic failure despite the use of otherwise appropriate and effective drugs. Frequently, infection in the immunocompetent host can be cured merely by halting multiplication of the microorganisms (a bacteriostatic effect). If host defenses are impaired, bacteriostatic activity may be inadequate and a bactericidal agent may be required for cure. Examples include bacterial endocarditis, where phagocytic cells are absent from the infected site; bacterial meningitis, where phagocytic cells are ineffective because of lack of opsonins; and disseminated bacterial infections in neutropenic patients, where the total mass of phagocytic cells is reduced. Patients with acquired immunodeficiency syndrome (AIDS) have impaired cellular immune responses, and therapy for various opportunistic infections in these patients is often suppressive but not curative. For example, most AIDS patients with bacteremia due to Salmonella will respond to conventional therapy, but this infection will relapse even after prolonged treatment (Jacobson et al., 1989), and treatment of disseminated atypical mycobacterial infection is lifelong.

Local Factors

Cure of an infection with antibiotics depends on an understanding of how local factors at the site of infection affect the antimicrobial activity of the drug. Antimicrobial activity may be significantly reduced in pus, which contains phagocytes, cellular debris, and proteins that can bind drugs or create conditions unfavorable to drug action (Bamberger et al., 1993; Konig et al., 1998). Large accumulations of hemoglobin in infected hematomas can bind penicillins and tetracyclines and may thus reduce the effectiveness of the drugs (Craig and Kunin, 1976). The pH in abscess cavities and in other confined infected sites (pleural space, CSF, and urine) is usually low, resulting in a marked loss of antimicrobial activity of aminoglycosides, erythromycin, and clindamycin (Strausbaugh and Sande, 1978). However, some drugs, such as chlortetracycline, nitrofurantoin, and methenamine, are more active in such an acidic environment. The anaerobic conditions found in abscess cavities impair activity of the aminoglycosides (Verklin and Mandell, 1977). Penetration of antimicrobial agents into infected areas such as abscess cavities is impaired, because the vascular supply is reduced. Successful therapy of abscesses usually requires drainage.

The presence of a foreign body in an infected site markedly reduces the likelihood of successful antimicrobial therapy. Prosthetic material [e.g., prosthetic cardiac valves, prosthetic joints, pacemakers, vascular grafts, and various vascular and central nervous system (CNS) shunts] is perceived by phagocytic cells as foreign. In an attempt to phagocytose and destroy it, degranulation occurs, resulting in the depletion of intracellular bactericidal substances. Thus, these phagocytes are relatively inefficient in killing bacterial pathogens. Bacteria also may reside within phagocytes, sequestered from most antimicrobial agents (Zimmerli et al., 1984). There also is evidence that bacteria adhering to prosthetic material resist killing by bactericidal agents (Chuard et al., 1991). Infections associated with foreign bodies are thus characterized by frequent relapses and failure, even with long-term, high-dose antibiotic therapy. Successful therapy usually requires removal of the foreign material.

Infectious agents that reside within phagocytic cells (intracellular parasites) are protected from the action of many antimicrobial agents that penetrate into cells poorly. This may be a problem in infections with Salmonella, Brucella, Toxoplasma, Listeria, and M. tuberculosis and, in some instances, even in infections caused by S. aureus. Certain antibioticsfor example, fluoroquinolones, isoniazid, trimethoprim-sulfamethoxazole, and rifampinpenetrate cells well and can achieve intracellular concentrations that inhibit or kill pathogens residing within cells.

Age

The age of the patient is an important determinant of pharmacokinetic properties of antimicrobial agents (see Chapter 1: Pharmacokinetics: The Dynamics of Drug Absorption, Distribution, and Elimination). Mechanisms of elimination, especially renal excretion and hepatic biotransformation, are poorly developed in the newborn, especially the premature infant. Failure to make adjustments for such differences can have disastrous consequences (e.g., see discussion of the 'gray-baby syndrome,' caused by chloramphenicol, in Chapter 47: Antimicrobial Agents: Protein Synthesis Inhibitors and Miscellaneous Antibacterial Agents). Elderly patients may clear renally eliminated drugs less well because of reduced creatinine clearance. They also may metabolize drugs less rapidly, predisposing them to elevated and potentially toxic concentrations of drugs when compared to younger patients. Elderly patients often are more likely to suffer toxicity at otherwise safe concentrations of drugs, as is the case for aminoglycoside ototoxicity.

Developmental factors also may determine the type of untoward response to a drug. Tetracyclines bind avidly to developing teeth and bones, and their use in young children can result in retardation of bone growth and discoloration or hypoplasia of tooth enamel. Fluoroquinolones accumulate in cartilage of developing bone, affecting its growth. Kernicterus may follow the use of sulfonamides in newborn infants, because this class of drugs competes effectively with bilirubin for binding sites on plasma albumin. Achlorhydria in young children and in the elderly or antacid therapy may alter absorption of orally administered antimicrobial agents (e.g., increased absorption of penicillin G and decreased absorption of ketoconazole).

Genetic Factors

Certain genetic or metabolic abnormalities must be considered when prescribing antibiotics. A number of drugs (e.g., sulfonamides, nitrofurantoin, chloramphenicol, and nalidixic acid) may produce acute hemolysis in patients with glucose-6-phosphate dehydrogenase deficiency. Patients who acetylate isoniazid rapidly may have subtherapeutic concentrations of the drug in plasma.

Pregnancy

Pregnancy may impose an increased risk of reaction to antimicrobial agents for both mother and fetus. Hearing loss in the child has been associated with administration of streptomycin to the mother during pregnancy. Tetracyclines can affect the bones and teeth of the fetus. Pregnant women receiving tetracycline may develop fatal acute fatty necrosis of the liver, pancreatitis, and associated renal damage. Pregnancy also may affect the pharmacokinetics of various antibiotics.

The lactating female can pass antimicrobial agents to her nursing child. Both nalidixic acid and sulfonamides in breast milk have been associated with hemolysis in children with glucose-6-phosphate dehydrogenase deficiency. In addition, sulfonamides, even in the small amounts received from breast milk, may predispose the nursing child to kernicterus (Vorherr, 1974).

Drug Allergy

Antibiotics, especially -lactams, are notorious for provoking allergic reactions. Patients with a history of atopic allergy seem particularly susceptible to the development of these reactions. The sulfonamides, trimethoprim, nitrofurantoin, and erythromycin also have been associated with hypersensitivity reactions, especially rash. A history of anaphylaxis (immediate reaction) or hives and laryngeal edema (accelerated reaction) precludes the use of the drug in all but extreme, life-threatening situations. Skin testing, particularly of the penicillins, may be of value in predicting life-threatening reactions. However, the utility of such tests is controversial. Antimicrobial agents, like other drugs, can cause 'drug fever,' which can be mistaken for a sign of continued infection.

Disorders of the Nervous System

Patients predisposed to seizures are at risk for localized or major motor seizures while taking high doses of penicillin G. This neurotoxicity of penicillin and other -lactam antibiotics correlates with high concentrations of drug in the CSF. It usually occurs in patients with renal insufficiency who are receiving large doses of these drugs. Decreased renal function increases concentrations of penicillin in CSF by two mechanisms: reduction of renal elimination of penicillin from plasma, which results in a higher concentration gradient for passive diffusion into CSF; and accumulation of organic acids (in the uremic state), which competitively inhibit the transport mechanism in the choroid plexus that removes penicillin and other organic acids from the CSF. Patients with myasthenia gravis or other neuromuscular problems are susceptible to the neuromuscular blocking effect of the aminoglycosides, polymyxins, and colistin. Patients undergoing general anesthesia who receive a neuromuscular blocking agent also are particularly susceptible to such antibiotic toxicity.

A large percentage (from 30% to 50%) of antimicrobial agents administered in the United States is given to prevent infection rather than to treat established disease. This practice accounts for some of the most flagrant misuses of these drugs.

Chemoprophylaxis is highly effective in some clinical settings, and in others is totally without value and may be deleterious. Use of antimicrobial compounds to prevent infections remains controversial in numerous situations. In general, if a single, effective, nontoxic drug is used to prevent infection by a specific microorganism or to eradicate an early infection, then chemoprophylaxis frequently is successful. On the other hand, if the aim of prophylaxis is to prevent colonization or infection by any or all microorganisms present in the environment of a patient, then prophylaxis often fails.

Prophylaxis may be used to protect healthy persons from acquisition of or invasion by specific microorganisms to which they are exposed. Successful examples of this practice include the following: use of rifampin to prevent meningococcal meningitis in people who are in close contact with a case; prevention of gonorrhea or syphilis after contact with an infected person; the intermittent use of trimethoprim-sulfamethoxazole to prevent recurrent urinary tract infections usually caused by E. coli.

Antimicrobial prophylaxis is used to prevent a variety of infections in patients undergoing organ transplantation or receiving cancer chemotherapy. Although specific infections often can be prevented, superinfections with opportunistic fungal pathogens or multiple-drug-resistant bacteria can be a problem. Moreover, the infection rate may be apparently lowered without changing overall outcomes.

Chemoprophylaxis is recommended for patients with valvular or other structural lesions of the heart predisposing to endocarditis who are undergoing dental, surgical, or other procedures that produce a high incidence of bacteremia (Dajani et al., 1997). Recent data suggesting that dental procedures have a minimal, if indeed any, role in causing endocarditis (Strom et al., 1998) have called some of the recommendations for chemoprophylaxis into question, but the recommendations nevertheless remain the standard of care. A procedure that injures a mucous membrane where there are large numbers of bacteria (such as in the oropharyngeal or gastrointestinal tract) will produce transient bacteremia. Streptococci from the mouth, enterococci from the gastrointestinal or genitourinary tract, and staphylococci from the skin commonly enter the bloodstream and may adhere to an abnormal or damaged valve surface, producing endocarditis. Chemoprophylaxis is directed against these microorganisms. Therapy, generally as a single dose, should not begin until immediately before the procedure, because prolonged administration of antibiotics can lead to colonization by resistant strains. Criteria have been established for the selection of specific drugs and patients who should receive chemoprophylaxis for various procedures.

The most extensive and probably best studied use of chemoprophylaxis is to prevent wound infections after various surgical procedures (see Table 432; see Antimicrobial prophylaxis in surgery, 1997). Wound infection results when a critical number of bacteria are present in the wound at the time of closure. Several factors determine the size of this critical inoculum, including virulence of the bacteria, the presence of devitalized or poorly vascularized tissue, the presence of a foreign body, and the status of the host. Antimicrobial agents directed against the invading microorganisms may reduce the number of viable bacteria below the critical level and thus prevent infection.

Several factors are important for the effective and judicious use of antibiotics for surgical prophylaxis. First, antimicrobial activity must be present at the wound site at the time of its closure. Thus, the drug should be given immediately preoperatively and perhaps intraoperatively for prolonged procedures. Second, the antibiotic must be active against the most likely contaminating microorganisms. Thus, cephalosporins are commonly used in this form of chemoprophylaxis. Third, there is mounting evidence that the continued use of drugs after the surgical procedure is unwarranted and potentially harmful. No data suggest that the incidence of wound infections is lower if antimicrobial treatment is continued after the day of surgery (Rowlands et al., 1982). Use beyond 24 hours not only is unnecessary, but also leads to the development of more resistant flora and superinfections caused by antibiotic-resistant strains. The risk of toxicity and the unnecessary expense are, of course, additional disadvantages.

Chemoprophylaxis should be limited to operative procedures for which there are data supporting its use. A number of studies indicate that it can be justified in dirty and contaminated surgical procedures (e.g., resection of the colon), where the incidence of wound infections is high. These include less than 10% of all surgical procedures. In clean surgical procedures, which account for approximately 75% of the total, the expected incidence of wound infection is less than 5%, and antibiotics should not be used routinely. Exceptions are rational when the surgery involves insertion of a prosthetic implant (e.g., prosthetic valve, vascular graft, prosthetic joint), cardiac surgery, or neurosurgical procedures; the complications of infection are so drastic that most authorities currently agree to chemoprophylaxis with these indications. Of course, the use of systemic antibiotics for chemoprophylaxis during surgical procedures does not reduce the need for sterile and skilled surgical technique.

All individuals who receive therapeutic doses of antibiotics undergo alterations in the normal microbial population of the intestinal, upper respiratory, and genitourinary tracts; some develop superinfection as a result of such changes. Superinfection may be defined as the appearance of bacteriological and clinical evidence of a new infection during the chemotherapy of a primary one. This phenomenon is relatively common and potentially very dangerous because the microorganisms responsible for the new infection are, in many cases, Enterobacteriaceae, Pseudomonas, and Candida or other fungi. These may be very difficult to eradicate with the currently available antiinfective drugs. Superinfection by these microorganisms is due to removal of the inhibitory influence of the normal flora. Normal flora may themselves produce antibacterial substances, and they also presumably compete for essential nutrients. The more 'broad' the effect of an antibiotic on microorganisms, the greater the alteration in the normal microflora and the greater the possibility that a single microorganism will become predominant, invade the host, and produce infection. Thus, the incidence of superinfection is lowest with penicillin G, higher with tetracyclines and chloramphenicol, and highest with combinations of broad-spectrum antimicrobials and the expanded-spectrum third-generation cephalosporins. It can be expected that further development and use of broad-spectrum agents will lead to more extensive alterations in the normal flora and, thus, to more superinfections. The development of agents that kill pathogens selectively while carefully preserving the normal flora would be beneficial. The most specific antimicrobial agent to treat a given infection should be chosen whenever possible. The incidence of superinfection also increases when administration of antibiotics is prolonged.

The fact that harmful effects may follow the therapeutic or prophylactic use of antiinfective agents should not discourage the physician from their administration in any situation in which they are definitely indicated. However, the physician should be hesitant to use antimicrobial drugs in instances where evidence of infection is entirely lacking or, at most, only suggestive. To do otherwise is to run the risk, at times, of converting a simple, benign, and self-limited disease into one that may be serious or even fatal.

The purpose of this introductory chapter has been to lay the groundwork for the maximally effective utilization of antimicrobial drugs. A brief discussion of the misuse and overuse of antimicrobial agents is in order.

Treatment of Untreatable Infections

A common misuse of these agents is in infections that have been proved by experimental and clinical observation to be untreatable: i.e., they do not respond to treatment with antimicrobial agents (Nyquist et al., 1998). The majority of the diseases caused by viruses are self-limited and do not respond to any of the currently available antiinfective compounds. Thus, antimicrobial therapy of measles, mumps, and at least 90% of infections of the upper respiratory tract and many gastrointestinal infections is ineffective and, therefore, useless.

Therapy of Fever of Unknown Origin

Fever of undetermined cause may be of two types: one that is present for only a few days to a week and another that persists for an extended period. Both of these are frequently (and inappropriately) treated with empirical antimicrobial agents. Fever of short duration, in the absence of localizing signs, is probably associated with undefined viral infections. Antimicrobial therapy is unnecessary and defervescence takes place spontaneously within a week or less. Fever persisting for two or more weeks, commonly referred to as 'fever of unknown origin,' has a variety of causes, of which only about one-quarter are infections (de Kleijn et al., 1997). Some of these infectionsfor example, tuberculosis or disseminated fungal infectionsmay require treatment with antimicrobial agents that are not used for common bacterial infections. Others, such as occult abscesses, may require surgical drainage or prolonged courses of pathogen-specific therapy, as in the case of bacterial endocarditis. Inappropriately administered antimicrobial therapy may serve to mask an underlying infection and contribute to delay in diagnosis, and prevent establishing a microbial etiology by rendering cultures negative. The noninfectious causesincluding regional enteritis, lymphoma, renal cell carcinoma, hepatitis, collagen-vascular disorders, and drug fever, to name a fewdo not respond to antimicrobial agents at all. Rather than embarking on a course of empirical antimicrobial therapy for fever of unknown origin, the physician should search for its cause.

Improper Dosage

Dosing errors, which can be the wrong frequency of administration or the use of either an excessive or a subtherapeutic dose, are common. Although antimicrobial drugs are among the safest and least toxic of drugs used in medical practice, excessive amounts can result in significant toxicities, including seizures (e.g., penicillin), vestibular damage (aminoglycosides), and renal failure (aminoglycosides), especially in patients with impaired drug excretion or metabolism. The use of too low a dose may result in treatment failure and is most likely to select for microbial resistance.

Inappropriate Reliance on Chemotherapy Alone

Infections complicated by abscess formation, the presence of necrotic tissue, or the presence of a foreign body often cannot be cured by antimicrobial therapy alone. Drainage, debridement, and removal of the foreign body are just as important as the choice of antimicrobial agent and in some cases more important. Two of many possible examples follow. The patient with pneumonia and empyema often fails to be cured even with administration of large doses of an effective drug unless the infected pleural fluid is drained. The patient with S. aureus bacteremia due to an intravascular device will continue to have fevers and positive blood cultures and be at an increased risk of dying unless the device is removed. As a general rule, when an appreciable quantity of pus, necrotic tissue, or a foreign body is present, the most effective treatment is an antimicrobial agent given in adequate dose plus a properly performed surgical procedure.

Lack of Adequate Bacteriological Information

One-half of the courses of antimicrobial therapy administered to hospitalized patients appear to be given in the absence of supporting microbiological data. Bacterial cultures and Gram stains of infected material are obtained too infrequently, and the results, when available, are often disregarded in the selection and application of drug therapy. Frequent use of drug combinations or drugs with the broadest spectra is a cover for diagnostic imprecision. The agents are selected more likely by habit than for specific indications, and the dosages employed are routine, rather than individualized on the basis of the clinical situation, microbiological information, and the pharmacological considerations presented in this and subsequent chapters of this section.

For general information regarding infectious diseases, the reader is referred to the following chapters of Harrison's Principles of Internal Medicine, 16th ed., McGraw-Hill, New York, 2005: basic considerations in infectious disease (Chapters 104 to 108), issues regarding bacterial infections (Chapter 118), viral diseases (Chapters 161 and 162), fungal diseases (Chapter 182), and parasitic infections (Chapters 192 and 193).