Pneumonia is an infection of the pulmonary parenchyma that can be caused by various bacterial species, including mycoplasmas, chlamydiae, and rickettsiae; viruses; fungi; and parasites (Table 255-1). Thus pneumonia is not a single disease but a group of specific infections, each with a different epidemiology, pathogenesis, clinical presentation, and clinical course. Identification of the etiologic microorganism is of primary importance, since this is the key to appropriate antimicrobial therapy. However, because of the serious nature of the infection, antimicrobial therapy generally needs to be started immediately, often before laboratory confirmation of the causative agent. The specific microbial etiology remains elusive in more than one-third of cases-e.g., when no sputum is available for examination, blood cultures are sterile, and there is no pleural fluid. Serologic confirmation requires weeks because of the late formation of specific antibody.

Table 255-1: Microbial Pathogens That Cause Pneumonia

Thus initial antimicrobial therapy is often empirical and is based on the setting in which the infection was acquired, the clinical presentation, patterns of abnormality on chest radiography, results of staining of sputum or other infected body fluids, and current patterns of susceptibility of the suspected pathogens to antimicrobial agents. After the etiologic agent is identified, specific antimicrobial therapy can be chosen.

The lung is a complex structure composed of aggregates of units that are formed by the progressive branching of the airways. Approximately 80% of the cells lining the central airways are ciliated, pseudostratified, columnar epithelial cells; the percentage decreases in the peripheral airways. Each ciliated cell contains about 200 cilia that beat in coordinated waves ~1000 times per minute, with a fast forward stroke and a slower backward recovery. Ciliary motion is also coordinated between adjacent cells so that each wave is propagated toward the oropharynx. The cilia are covered by a liquid film that is ~5 to 10 m thick and is composed of two layers. The outer, or gel, layer is viscous and traps deposited particles. The cilia beat in the less viscous inner, or sol, layer. During the forward stroke, the tips of the cilia just touch the viscous gel and propel it toward the oropharynx. During recovery, the cilia move entirely within the low-resistance sol layer. Ciliated cells are interspersed with mucus-secreting cells in the trachea and bronchi but not in the bronchioles.

The alveolar walls, from blood to air, consist of the endothelium that lines the network of anastomotic capillaries, the capillary basement membrane, the interstitial tissue, the alveolar basement membrane, the alveolar lining epithelial cells (which are either flattened type I pneumocytes that cover 95% of the alveolar surface or rounded, granular, surfactant-producing type II pneumocytes), and epithelial lining fluid. The epithelial lining fluid contains surfactant, fibronectin, and immunoglobulin, which may opsonize or-in the presence of complement-lyse microbial pathogens deposited on the alveolar surface. Loosely attached to the lining cells or lying free within the lumen are the alveolar macrophages, lymphocytes, and a few polymorphonuclear leukocytes.

The lower respiratory tract is normally sterile, despite being adjacent to enormous numbers of microorganisms that reside in the oropharynx and being exposed to environmental microorganisms in inhaled air. This sterility is the result of efficient filtering and clearance mechanisms.

Infectious particles deposited on the squamous epithelium of distal nasal surfaces normally are removed by sneezing, while those deposited on the more proximal ciliated surfaces are swept posteriorly in the mucus lining into the nasopharynx, where they are swallowed or expectorated. Reflex closure of the glottis and cough protect the lower respiratory tract. Those particles deposited on the tracheobronchial surface are swept by ciliary motion toward the oropharynx. Infectious particles that bypass defenses in the airways and are deposited on the alveolar surface are cleared by phagocytic cells and humoral factors. Alveolar macrophages are the major phagocytes in the lower respiratory tract. Some phagocytosed microorganisms are killed by the phagocyte's oxygen-dependent systems, lysosomal enzymes, and cationic proteins. Other microorganisms can evade microbicidal mechanisms and persist within the macrophage. For example, Mycobacterium tuberculosis persists within the lysosome, while Legionella resides within intracellular inclusions that fail to fuse with lysosomes. Intracellular pathogens can then be transported to the ciliated surfaces and into the oropharynx or via the lymphatics to regional lymph nodes. The alveolar macrophages process and present microbial antigens to the lymphocyte and also secrete cytokines (e.g., tumor necrosis factor and interleukin 1) that modulate the immune process in T and B lymphocytes. Cytokines facilitate the generation of an inflammatory response, activate alveolar macrophages, and recruit additional phagocytes and other immunologic factors from plasma. The inflammatory exudate is responsible for many of the local signs of pulmonary consolidation and for the systemic manifestations of pneumonia, such as fever, chills, myalgias, and malaise.

Microbial pathogens may enter the lung by one of several routes.

Most pulmonary pathogens originate in the oropharyngeal flora. Aspiration of these pathogens is the most common mechanism for the production of pneumonia. At various times during the year, healthy individuals transiently carry common pulmonary pathogens in the nasopharynx; these pathogens include Streptococcus pneumoniae, S. pyogenes, Mycoplasma pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis. The sources of anaerobic pulmonary pathogens, such as Porphyromonas gingivalis, Prevotella melaninogenica, Fusobacterium nucleatum, Actinomyces spp., spirochetes, and anaerobic streptococci, are the gingival crevice and dental plaque, which contain more than 10¹¹ colony-forming units (CFU) of microorganisms per gram. The frequency of aerobic gram-negative bacillary colonization of the oropharyngeal mucosa, which is unusual in healthy persons (<2%), increases with hospitalization, worsening debility, severe underlying illness, alcoholism, diabetes, and advanced age. This change may be a consequence of increased salivary proteolytic activity, which destroys fibronectin, a glycoprotein coating the surface of the mucosa. Fibronectin is the receptor for the normal gram-positive flora of the oropharynx. Loss of fibronectin exposes the receptors for aerobic gram-negative bacilli on the epithelial cell surface. The source of aerobic gram-negative bacilli may be the patient's own stomach (which can become colonized with these organisms as the result of an increase in gastric pH with atrophic gastritis or after the use of H₂-blocking agents or antacids), contaminated respiratory equipment, hands of health care workers, or contaminated food and water. Nasogastric tubes can facilitate the transfer of gastric bacteria to the pharynx.

About 50% of healthy adults aspirate oropharyngeal secretions into the lower respiratory tract during sleep. Aspiration occurs more frequently and may be more pronounced in individuals with an impaired level of consciousness (e.g., alcoholics; drug abusers; and patients who have had seizures, strokes, or general anesthesia), neurologic dysfunction of the oropharynx, and swallowing disorders or mechanical impediments (e.g., nasogastric or endotracheal tubes). Pneumonia due to anaerobes is an especially likely outcome if the aspirated material is large in volume or contains virulent components of the anaerobic microbial flora or foreign bodies, such as aspirated food or necrotic tissue. Impairment of the cough reflex increases the risk of pneumonia, as does mucociliary or alveolar macrophage dysfunction.

Deposition of inhaled particles within the respiratory tract is determined primarily by particle size. Particles >10 m m in diameter are deposited mostly in the nose and upper airways. Particles <5 m m in diameter (also called airborne droplet nuclei) and containing one or perhaps two microorganisms fail to settle out by gravity but rather remain suspended in the atmosphere for long periods unless removed by ventilation or by filtration in the lungs of the individual breathing the contaminated air. Transmission of an infectious agent in the form of an aerosol is particularly efficient. These infectious aerosols are small enough to bypass host defenses in the upper respiratory tract and airways. A greater percentage of particles are deposited in small bronchioles and alveoli as particle size decreases below 5 m m. One inhaled particle of appropriate size may be sufficient to reach the alveolus and initiate infection. The etiologies of pneumonia typically acquired by inhalation of infectious aerosols include tuberculosis, influenza, legionellosis, psittacosis, histoplasmosis, Q fever, and hantavirus pulmonary syndrome (HPS).

Infection, usually with Staphylococcus aureus, disseminates hematogenously to the lungs in patients (such as intravenous drug users) who have either right- or left-sided bacterial endocarditis and in patients with intravenous catheter infections. Fusobacterium infections of the retropharyngeal tissues (Lemierre's syndrome-i.e., retropharyngeal abscess and jugular venous thrombophlebitis) also disseminate hematogenously to the lungs.

Two additional routes of transmission of bacteria to the lungs are direct inoculation (as a result of either tracheal intubation or stab wounds to the chest) and contiguous spread from an adjacent site of infection.

The pneumonic process may involve primarily the interstitium or the alveoli. Involvement of an entire lobe is called lobar pneumonia. When the process is restricted to alveoli contiguous to bronchi, it is called bronchopneumonia. Confluent bronchopneumonia may be indistinguishable from lobar pneumonia. Cavities develop when necrotic lung tissue is discharged into communicating airways, resulting in either necrotizing pneumonia (multiple small cavities, each <2 cm in diameter, in one or more bronchopulmonary segments or lobes) or lung abscess (one or more cavities >2 cm in diameter). The classification of pneumonia is best based upon the causative microorganism rather than upon these anatomic characteristics (the criteria used in the past).

The patient's living circumstances, occupation, travel history, pet or animal exposure history, and contacts with other ill individuals as well as the physician's knowledge of the epidemic curve of community outbreaks provide clues to the microbial etiology of a given case of pneumonia (Table 255-1). The relative frequency of various pulmonary pathogens varies with the setting in which the infection was acquired-e.g., community, nursing home, or hospital. In patients hospitalized with community-acquired pneumonia, the most frequent pathogens are S. pneumoniae, H. influenzae, Chlamydia pneumoniae, and Legionella pneumophila. C. pneumoniae is often found in association with other pathogens, including S. pneumoniae, and the associated pathogen appears to influence the course of the pneumonia. M. pneumoniae, which usually causes mild illness, is common among outpatients with community-acquired pneumonia, but may also be an underappreciated cause in all age groups of severe pneumonia that requires hospitalization. In contrast, enteric aerobic gram-negative bacilli and Pseudomonas aeruginosa, uncommon causes of community-acquired pneumonia, are estimated to account for >50% of cases of hospital-acquired pneumonia, while S. aureus is responsible for >10%. The relative frequencies of pathogens in pneumonia acquired in nursing homes fall somewhere between those of community- and hospital-acquired pneumonia. Enteric aerobic gram-negative bacilli and P. aeruginosa are more common among nursing home residents than among patients who acquire pneumonia in noninstitutional settings.

The season of the year and the geographic location are other predictors of etiology. The frequency of influenza virus as a cause of both community-acquired and institutionally acquired pneumonia increases during the winter months. Moreover, influenza virus infection causes an increase in the frequency of secondary bacterial pneumonia due to S. pneumoniae, S. aureus, and H. influenzae. Outbreaks of influenza in a community tend to be explosive and widespread, with many secondary cases resulting from the short incubation period of several days and the high degree of communicability. Legionella colonizes hot-water storage systems that provide favorable conditions for its proliferation, such as warm temperature, stagnation, and sediment accumulation. Acquisition of Legionella pneumonia requires exposure to aerosols generated from these contaminated water supplies-e.g., during an overnight stay in a hotel with a faulty air-handling system or after repair of domestic plumbing in buildings with contaminated water supplies. Legionellosis also occurs in explosive outbreaks when large numbers of susceptible people are exposed to an infectious aerosol; however, no secondary cases occur because of the low level of communicability of L. pneumophila. Mycoplasma causes outbreaks, usually in relatively closed populations such as those at military bases, at colleges, or in households; however, because of its long incubation period (2 to 3 weeks) and its relatively low degree of communicability, Mycoplasma infection moves through the community slowly, affecting another person as the first is recovering. In communities where infection with HIV type 1 is endemic, Pneumocystis carinii and M. tuberculosis are more prominent causes of community-acquired pneumonia. Chlamydia psittaci produces illness in bird handlers. Histoplasmosis, blastomycosis, and coccidioidomycosis are causes of pneumonia that have specific geographic distributions.

HPS is a newly described, frequently fatal disease caused by one of several hantaviruses. Most cases in the United States have been reported from the Four Corners area (New Mexico, Arizona, Utah, and Colorado), where the pathogen is the Sin Nombre virus. The primary hosts are rodents, which apparently remain healthy but excrete the virus in urine, feces, and saliva. Hantavirus infection is acquired by inhalation of infectious aerosols when rodent nests are disturbed by human domestic, occupational, or recreational activities. The appearance of HPS in the southwestern United States is thought to have occurred because of increased rainfall in the region, which increased the rodent food supply and thus the rodent population. No person-to-person transmission of HPS is thought to have taken place, except perhaps in an outbreak in southern Argentina in 1996.

Age is an important predictor of the infecting agent in pneumonia. Chlamydia trachomatis and respiratory syncytial virus are common among infants < 6 months of age; H. influenzae among children 6 months to 5 years of age; M. pneumoniae, C. pneumoniae, and hantavirus among young adults; H. influenzae and M. catarrhalis among elderly individuals with chronic lung disease; and L. pneumophila among elderly persons, smokers, and persons with compromised cell-mediated immunity (e.g., transplant recipients), renal or hepatic failure, diabetes, or systemic malignancy.

Oral anaerobes, frequently in combination with aerobic bacterial components of the human flora (e.g., viridans streptococci), are causes of community-acquired pneumonia and anaerobic lung abscess in patients who are prone to aspiration. Edentulous persons, who have lower numbers of oral anaerobes, are less likely to develop pneumonia due to anaerobes. When the etiology of community-acquired pneumonia in unselected hospitalized patients has been studied by methods that entail strict anaerobic bacteriology and that avoid contamination of lower respiratory tract secretions by the oral flora, anaerobic bacteria have been found to account for as many as 20 to 30% of cases. In hospital-acquired pneumonia, anaerobes are the pathogens-with or without aerobic copathogens-in about one-third of cases. However, the aerobic copathogens in hospital-acquired pneumonia are frequently virulent microorganisms in their own right (e.g., enteric aerobic gram-negative bacilli, P. aeruginosa, and S. aureus).

The patient's underlying disease may be characterized by specific immunologic or inflammatory defects that predispose to pneumonia due to specific pathogens (Table 255-2). For example, immunoglobulin deficiencies-especially those involving IgG subtypes 2 and 4, which are important in the immune response to encapsulated organisms (e.g., S. pneumoniae and H. influenzae)-may be associated with recurrent sinopulmonary infections. Immunoglobulin deficiencies may be inherited, or they may be acquired (i.e., as a result of either decreased production, as in lymphoproliferative malignancies, or excessive protein loss, as in nephrosis or protein-losing enteropathy). Inherited immunoglobulin deficiencies may be global or selective. Patients with recurrent sinopulmonary infections and a selective deficiency of IgG2 and/or IgG4 may have a total plasma IgG level within the normal range, as these particular IgG subtypes constitute only 25% of total IgG. HIV-infected patients may also exhibit ineffective antibody formation, which predisposes to infection with these encapsulated bacteria. Severe neutropenia (<500 neutrophils/m L) increases the risk of infections due to P. aeruginosa, Enterobacteriaceae, S. aureus, and (if neutropenia is prolonged) Aspergillus. The risk is unusually high for infections due to M. tuberculosis among HIV-infected patients with circulating CD4+ lymphocyte counts of <500/m L; for infections due to P. carinii, Histoplasma capsulatum, and Cryptococcus neoformans among those with CD4+ counts of <200/m L; and for infections due to M. avium-intracellulare and cytomegalovirus among those with counts of <50/m L. Long-term glucocorticoid therapy increases the risk of tuberculosis and nocardiosis.

Table 255-2: Pulmonary Pathogens Associated with Specific Defects in Host Defenses

Community-acquired pneumonia has traditionally been thought to present as either of two syndromes: the typical presentation or the atypical presentation. Although current data suggest that these two syndromes may be less distinct than was once thought, the characteristics of the clinical presentation may nevertheless have some diagnostic value.

The "typical" pneumonia syndrome is characterized by the sudden onset of fever, cough productive of purulent sputum, shortness of breath, and (in some cases) pleuritic chest pain; signs of pulmonary consolidation (dullness, increased fremitus, egophony, bronchial breath sounds, and rales) may be found on physical examination in areas of radiographic abnormality. The typical pneumonia syndrome is usually caused by the most common bacterial pathogen in community-acquired pneumonia, S. pneumoniae, but can also be due to other bacterial pathogens, such as H. influenzae and mixed anaerobic and aerobic components of the oral flora.

The "atypical" pneumonia syndrome is characterized by a more gradual onset, a dry cough, shortness of breath, a prominence of extrapulmonary symptoms (such as headache, myalgias, fatigue, sore throat, nausea, vomiting, and diarrhea), and abnormalities on chest radiographs despite minimal signs of pulmonary involvement (other than rales) on physical examination. Atypical pneumonia is classically produced by M. pneumoniae but can also be caused by L. pneumophila, C. pneumoniae, oral anaerobes, and P. carinii as well as by S. pneumoniae and the less frequently encountered pathogens C. psittaci, Coxiella burnetii, Francisella tularensis, H. capsulatum, and Coccidioides immitis. Mycoplasma pneumonia (Chap. 178) may be complicated by erythema multiforme, hemolytic anemia, bullous myringitis, encephalitis, and transverse myelitis. Legionella pneumonia (Chap. 151) is frequently associated with deterioration in mental status, renal and hepatic abnormalities, and marked hyponatremia; pneumonia due to H. capsulatum (Chap. 201) or C. immitis (Chap. 202) is often accompanied by erythema nodosum. In C. pneumoniae pneumonia (Chap. 179), sore throat, hoarseness, and wheezing are relatively common. The atypical pneumonia syndrome in patients whose behavioral history places them at risk of HIV infection suggests Pneumocystis infection. These patients may have concurrent infections caused by other opportunistic pathogens, such as pulmonary (and frequently extrapulmonary) tuberculosis, oral thrush due to Candida albicans, or extensive perineal ulcers due to herpes simplex virus.

Certain viruses also produce pneumonia that is usually characterized by an atypical presentation-i.e., chills, fever, shortness of breath, dry nonproductive cough, and predominance of extrapulmonary symptoms. Primary viral pneumonia can be caused by influenza virus (usually as part of a community outbreak in winter), by respiratory syncytial virus (in children and immunosuppressed individuals), by measles or varicella-zoster virus (accompanied by the characteristic rash), and by cytomegalovirus (in patients immunocompromised by HIV infection or by therapy given in association with organ transplantation). Hantavirus causes an initial nonspecific febrile prodrome, after which the patient develops rapidly progressive respiratory failure and diffuse pulmonary infiltrates on chest radiographs as a result of exudation into the pulmonary interstitium and alveoli, with thrombocytopenia, neutrophilic leukocytosis, circulating immunoblasts, and laboratory evidence of hemoconcentration. In addition, influenza and measles can predispose to secondary bacterial pneumonia as a result of the destruction of the mucociliary barrier of the airways. Secondary bacterial infection may either follow the viral infection without interruption or be separated from the viral infection by several days of transient relief of symptoms. Bacterial infection may be heralded by sudden worsening of the patient's clinical condition, with persisting or renewed chills, fever, and cough productive of purulent sputum, possibly accompanied by pleuritic chest pain.

Patients with hematogenous S. aureus pneumonia may present with fever and dyspnea only. In these cases the inflammatory response is initially confined to the pulmonary interstitium. Cough, sputum production, and signs of pulmonary consolidation develop only after the infection extends into the bronchi. These patients are usually gravely ill, with intravascular infection as well as pneumonia, and may have signs of endocarditis (Chap. 126).

Nocardiosis (Chap. 165) is frequently complicated by metastasis of lesions to the skin and central nervous system. Signs of pulmonary consolidation, cough, and sputum production may be lacking in patients who are unable to mount an inflammatory response, such as those with agranulocytosis. The major manifestations in these patients may be limited to fever, tachypnea, agitation, and altered mental status. Elderly or severely ill patients may fail to develop fever.

Tuberculosis also produces an atypical presentation that is characterized by fever, night sweats, cough, and shortness of breath and sometimes by pleuritic chest pain and blood-streaked sputum. Several weeks usually elapse before the patient seeks medical attention because of the gradual worsening of these symptoms, by which time he or she will have lost considerable weight.

Patients with nosocomial pneumonia often pose a diagnostic challenge. The differential diagnosis of acute respiratory disease in critically ill, hospitalized patients is diverse and includes noninfectious entities, such as congestive heart failure, acute respiratory distress syndrome, preexisting lung disease, atelectasis, and oxygen- or drug-related toxicities, that may be difficult to distinguish clinically or radiologically from pneumonia. The usual criteria for nosocomial pneumonia, which include new or progressive pulmonary infiltrates, purulent tracheobronchial secretions, fever, and leukocytosis, are frequently unreliable in these patients, who often have preexisting pulmonary disease, endotracheal tubes that irritate the tracheal mucosa and may elicit an inflammatory exudate in respiratory secretions, or multiple other problems likely to produce fever and leukocytosis. Patients with nosocomial pneumonia complicating an underlying illness associated with significant neutropenia often have no purulent respiratory tract secretions or pulmonary infiltrates, and patients with nosocomial pneumonia complicating uremia or cirrhosis often remain afebrile. In addition, the patients at greatest risk for nosocomial pneumonia are most likely to be heavily colonized with potential pulmonary pathogens in the oropharyngeal or tracheobronchial mucosa; thus the presence of these organisms in gram-stained preparations or cultures of respiratory tract secretions does not necessarily confirm the diagnosis of pneumonia.

Aspiration of a sufficient volume of gastric acid produces a chemical pneumonitis characterized by acute dyspnea and wheezing with hypoxemia and infiltrates on chest radiographs in one or both lower lobes. Clinical findings following aspiration of particulate matter depend on the extent of endobronchial obstruction and range from acute apnea to persistent cough with or without recurrent infection. Although the aspiration of oral anaerobes can initially lead to an infiltrative process, it ultimately results in putrid sputum, tissue necrosis, and pulmonary cavities. In about three-quarters of cases, the clinical course of an abscess of anaerobic polymicrobial etiology is indolent and mimics that of pulmonary tuberculosis, with cough, shortness of breath, chills, fever, night sweats, weight loss, pleuritic chest pain, and blood-streaked sputum lasting for several weeks or more. In other patients the disease may present more acutely. Patients with anaerobic abscesses are usually prone to aspiration of oropharyngeal contents and have periodontal disease. One genus of oral anaerobes, Actinomyces, produces a chronic fibrotic necrotizing process that crosses tissue planes and may involve the pleural space, ribs, vertebrae, and subcutaneous tissue, with eventual discharge of sulfur granules (macroscopic bacterial masses) through the skin (empyema necessitatis).

Chest radiography is more sensitive than physical examination for detection of pulmonary infiltrates. Indeed, P. carinii pneumonia (PCP) is the only relatively common form of pneumonia associated with false-negative chest radiographs; up to 30% of patients with PCP have false-negative results. Chest radiographs can confirm the presence and location of the pulmonary infiltrate; assess the extent of the pulmonary infection; detect pleural involvement, pulmonary cavitation, or hilar lymphadenopathy; and gauge the response to antimicrobial therapy. However, chest radiographs may be normal when the patient is unable to mount an inflammatory response (e.g., in agranulocytosis) or is in the early stage of an infiltrative process (e.g., in hematogenous S. aureus pneumonia or PCP associated with AIDS). High-resolution computed tomography of the lungs can improve the accuracy of diagnosis of pneumonia, especially when the process involves lung obscured by the diaphragm, liver, ribs and clavicles, or heart.

The anatomic localization of the inflammatory process, as visualized in chest radiographs, occasionally has diagnostic implications. Most pulmonary pathogens produce focal lesions. A multicentric distribution suggests hematogenous infection, in which case the remote location of the primary infection (e.g., endocarditis or thrombophlebitis) should be sought. Hematogenous pneumonia, which results from septic embolization in patients with thrombophlebitis or right-sided endocarditis and from bacteremia in patients with left-sided endocarditis, appears on the chest radiograph as multiple areas of pulmonary infiltration that subsequently may cavitate. A diffuse distribution suggests the involvement of P. carinii, cytomegalovirus, hantavirus, measles virus, or herpes zoster virus (with pneumonia due to the last two pathogens diagnosed by the characteristic accompanying rash). Pleurisy and hilar nodal enlargement are unusual with PCP and cytomegalovirus pneumonia; their presence suggests another etiology. Diffuse lesions in immunocompromised patients also suggest legionellosis, tuberculosis, histoplasmosis, Mycoplasma infection, or disseminated strongyloidiasis.

Oral anaerobes, S. aureus, S. pneumoniae serotype III, aerobic gram-negative bacilli, M. tuberculosis, and fungi as well as certain noninfectious conditions can produce tissue necrosis and pulmonary cavities (Table 255-3). In contrast, H. influenzae, M. pneumoniae, viruses, and most other serotypes of S. pneumoniae almost never cause cavities. Apical disease, with or without cavities, suggests reactivation tuberculosis. Anaerobic abscesses are located in dependent, poorly ventilated, and poorly draining bronchopulmonary segments and characteristically have air-fluid levels, unlike the well-ventilated, well-drained upper-lobe cavities caused by M. tuberculosis, an obligate aerobe. Air-fluid levels may also be present in cavities due to pulmonary necrosis of other infectious etiologies, such as S. aureus and aerobic gram-negative bacilli. Mucor and Aspergillus invade blood vessels and cause pleural-based, wedge-shaped areas of pulmonary infarction; these infarcts may subsequently cavitate.

Table 255-3: Causes of Pulmonary Cavities

In the patient with an uncomplicated course, chest radiographs need not be repeated before discharge, since the resolution of infiltrates may take up to 6 weeks after initial presentation. However, patients who do not respond clinically, who have a pleural effusion on admission, who may have postobstructive pneumonia, or who are infected with certain pathogens (e.g., S. aureus, aerobic gram-negative bacilli, or oral anaerobes) need more intensive surveillance. At times, computed tomography may be especially helpful in distinguishing different processes-e.g., pleural effusion versus underlying pulmonary consolidation, hilar adenopathy versus pulmonary mass, and pulmonary abscess versus empyema with an air-fluid level.

Examination of the sputum remains the mainstay of the evaluation of a patient with acute bacterial pneumonia. Unfortunately, expectorated material is frequently contaminated by potentially pathogenic bacteria that colonize the upper respiratory tract (and sometimes the lower respiratory tract) without actually causing disease. This contamination reduces the diagnostic specificity of any lower respiratory tract specimen. In addition, it has been estimated that the usual laboratory processing methods detect the pulmonary pathogen in fewer than 50% of expectorated sputum samples from patients with bacteremic S. pneumoniae pneumonia. This low sensitivity may be due to misidentification of the b -hemolytic colonies of S. pneumoniae as nonpathogenic b -hemolytic streptococci ("normal flora"), overgrowth of the cultures by hardier colonizing organisms, or loss of more fastidious organisms due to slow transport or improper processing. In addition, certain common pulmonary pathogens, such as anaerobes, mycoplasmas, chlamydiae, Pneumocystis, mycobacteria, fungi, and legionellae, cannot be cultured by routine methods.

Since expectorated material is routinely contaminated by oral anaerobes, the diagnosis of anaerobic pulmonary infection is frequently inferred. Confirmation of such a diagnosis requires the culture of anaerobes from pulmonary secretions that are uncontaminated by oropharyngeal secretions, which in turn requires the collection of pulmonary secretions by special techniques, such as transtracheal aspiration (TTA), transthoracic lung puncture, and protected brush via bronchoscopy. These procedures are invasive and are usually not used unless the patient fails to respond to empirical therapy.

Gram's staining of sputum specimens, screened initially under low-power magnification (10× objective and 10× eyepiece) to determine the degree of contamination with squamous epithelial cells, is of utmost diagnostic importance. In patients with the typical pneumonia syndrome who produce purulent sputum, the sensitivity and specificity of Gram's staining of sputum minimally contaminated by upper respiratory tract secretions (>25 polymorphonuclear leukocytes and <10 epithelial cells per low-power field) in identifying the pathogen as S. pneumoniae are 62 and 85%, respectively. Gram's staining in this case is more specific and probably more sensitive than the accompanying sputum culture. The finding of mixed flora on Gram's staining of an uncontaminated sputum specimen suggests an anaerobic infection. Acid-fast staining of sputum should be undertaken when mycobacterial infection is suspected. Examination by an experienced pathologist of Giemsa-stained expectorated respiratory secretions from patients with AIDS has given satisfactory results in the diagnosis of PCP. The sensitivity of sputum examination is enhanced by the use of monoclonal antibodies to Pneumocystis and is diminished by prior prophylactic use of inhaled pentamidine. Blastomycosis can be diagnosed by the examination of wet preparations of sputum. Sputum stained directly with fluorescent antibody can be examined for Legionella, but this test yields false-negative results relatively often. Thus sputum should also be cultured for Legionella on special media.

Expectorated sputum usually is easily collected from patients with a vigorous cough but may be scant in patients with an atypical syndrome, in the elderly, and in persons with altered mental status. If the patient is not producing sputum and can cooperate, respiratory secretions should be induced with ultrasonic nebulization of 3% saline. An attempt to obtain lower respiratory secretions by passage of a catheter through the nose or mouth rarely achieves the desired results in an alert patient and is discouraged; usually the catheter can be found coiled in the oropharynx.

In some cases that do not require the patient's hospitalization (see "Decision to Hospitalize," below), an accurate microbial diagnosis may not be crucial, and empirical therapy can be started on the basis of clinical and epidemiologic evidence alone. This approach may also be appropriate for hospitalized patients who are not severely ill and who are unable to produce an induced sputum specimen. Use of invasive procedures to establish a microbial diagnosis carries risks that must be weighed against potential benefits. However, the decision to initiate empirical therapy without an evaluation of induced sputum should be undertaken with caution and, in the case of hospitalized patients, should always be accompanied by the culture of several blood samples. The ability to understand the cause of a poor response to empirical antimicrobial therapy (Table 255-4) may be compromised by the lack of initial sputum and blood cultures. Establishing a specific microbial etiology in the individual patient is important, for it allows institution of specific pathogen-directed antimicrobial therapy and reduces the use of broad-spectrum combination regimens to cover multiple possible pathogens. Use of a single narrow-spectrum antimicrobial agent exposes the patient to fewer potential adverse drug reactions and reduces the pressure for selection of antimicrobial resistance. Emergence of antimicrobial resistance is a type of adverse drug reaction unlike others, because it is "contagious." In addition, establishing a microbial diagnosis can help define local community outbreaks and antimicrobial resistance patterns.

Table 255-4: Factors Involved in Poor Response to Empirical Antimicrobial Therapy

The sensitivities and specificities of the invasive procedures described below for obtaining pulmonary material vary with the type of immunocompromised patient, the type of pulmonary lesion, and the degree of prior exposure to therapeutic or prophylactic antimicrobial agents.

Popular several decades ago, TTA is rarely performed today. Although the sensitivity of the procedure is high (approaching 90%), the specificity is low. The material obtained by TTA (from a catheter inserted through the cricothyroid cartilage and advanced toward the carina) is not contaminated by upper respiratory tract secretions but can contain organisms that colonize the tracheobronchial tree without necessarily causing pneumonia. Significant morbidity and even death have attended the use of TTA. Contraindicated in patients with a bleeding diathesis, TTA may cause infection at the puncture site and may lead to severe subcutaneous and mediastinal emphysema in patients who are coughing vigorously.

This procedure employs a skinny (small-gauge) needle that is advanced into the area of pulmonary consolidation with computed tomographic guidance. It requires that the patient cooperate, have good hemostasis, and be able to tolerate a possible associated pulmonary hemorrhage or pneumothorax. Patients on mechanical ventilation cannot undergo lung puncture because of the high incidence of complicating pneumothorax.

Fiberoptic bronchoscopy is safe and relatively well tolerated and has become the standard invasive procedure used to obtain lower respiratory tract secretions from seriously ill or immunocompromised patients with complex or progressive pneumonia. This technique provides a direct view of the lower airways. Specimens obtained by bronchoscopy should be subjected to Gram's, acid-fast, Legionella direct fluorescent antibody, and Gomori's methenamine silver staining and should be cultured for routine aerobic and anaerobic bacteria, legionellae, mycobacteria, and fungi. Samples are collected with a protected double-sheathed brush (PSB), by bronchoalveolar lavage (BAL), or by transbronchial biopsy (TBB) at the site of pulmonary consolidation. The PSB sample is usually contaminated by oropharyngeal flora; quantitative cultures of the 1 mL of sterile culture medium into which the brush is placed after withdrawal from the inner catheter must be performed to differentiate contamination (<1000 CFU/mL) from infection (1000 CFU/mL). The results of PSB are highly specific and highly sensitive, especially when the patient has not received antibiotics before culture. BAL is usually performed with 150 to 200 mL of sterile, nonbacteriostatic saline. When used to facilitate endoscopy, local anesthetic agents with antibacterial activity can lower the sensitivity of culture results. Quantitative bacteriologic evaluation of BAL fluid has given results similar to those obtained with the PSB technique. Gram's staining of the cytocentrifuged BAL fluid specimen can serve as an immediate guide in the selection of antimicrobial therapy to be administered while culture results are awaited.

This procedure is most commonly needed when specimens obtained bronchoscopically from an immunocompromised patient with progressive pneumonia have been unrevealing. Limitations on the performance of an open-lung biopsy include hypoxemia and a bleeding diathesis, which may supervene while the physician is deciding whether to undertake this procedure. Results of an open-lung biopsy are considered diagnostic because of the large size of the tissue sample. The diagnostic yield of this procedure is greatest in focal lesions, whereas bronchoscopic evaluation is most useful in diffuse lesions.

In the initial evaluation of a patient with pneumonia, at least two blood samples for culture should be obtained from different venipuncture sites; if empyema is a clinical consideration, diagnostic thoracentesis is indicated. Positive blood or pleural fluid culture is generally considered diagnostic of the etiology of pneumonia. However, bacteremia and empyema each occur in fewer than 10 to 30% of patients with pneumonia.

Serologic studies are sometimes helpful in defining the etiology of certain types of pneumonia, although serologic diagnosis-because it is often delayed by the need to demonstrate at least a fourfold rise in convalescent-phase antibody titer-is usually retrospective. A single IgM antibody titer of >1:16, a single IgG antibody titer of >1:128, or a fourfold or greater rise in the IgG titer obtained by indirect immunofluorescence is diagnostic of M. pneumoniae infection. A single IgM antibody titer of 1:20, a single IgG antibody titer of 1:128, or a fourfold or greater rise in the IgG titer obtained by micro-indirect immunofluorescence is diagnostic of C. pneumoniae infection. A single Legionella antibody titer of 1:256 or a fourfold rise to a titer of 1:128 suggests acute legionellosis. A highly sensitive and specific urinary antigen test is available to detect L. pneumophila serogroup 1 in patients with pneumonia; this organism accounts for ~70% of L. pneumophila infections. The diagnosis of hantavirus infection is confirmed by detection of IgM serum antibodies, a rising titer of IgG serum antibodies, hantavirus-specific RNA by polymerase chain reaction in clinical specimens, and hantavirus-specific antigen by immunohistochemistry.

Approximately 20% of patients with community-acquired pneumonia are hospitalized, some perhaps unnecessarily. Use of inpatient hospital services is costly and at times poses risks to the patient (e.g., the risk of nosocomial infections). Thus hospitalization must be justified by anticipation of a poor outcome if the case is managed in an outpatient setting.

The Pneumonia Patient Outcomes Research Team (PORT) has attempted to quantify the risk of death and other adverse outcomes of community-acquired pneumonia by assignment of points to 19 variables (Fig. 255-1), with stratification of patients into five classes based on cumulative point score. This prediction rule was derived and validated in a large number of patients. On the basis of their observations, the PORT investigators suggest that outpatient management is appropriate for many patients in classes I and II, in whom the risks of subsequent hospitalization (8.2%) and of death (<0.6%) are low. They suggest outpatient management after a short hospital stay for patients in class III, whose risk of subsequent hospitalization if initially treated at home is 16.7% but whose risk of admission to the intensive care unit (ICU) is 5.9%-similar to that for patients in classes I and II. The PORT investigators further suggest that patients in classes IV and V (risk of death, 8.2 and 29.2%, respectively; risk of ICU admission, 11.4 and 17.3%, respectively) should receive traditional inpatient care. An expert panel from the Infectious Diseases Society of America (IDSA) endorses the PORT recommendations.

Figure 255-1: Criteria for hospitalization of patients with pneumonia: the PORT score. *A risk score (total point score) for a given patient is obtained by summing the patient's age in years (age minus 10 for females) and the points for each applicable patient characteristic. Oxygen saturation of <90% is also considered abnormal.

Other characteristics that favor a decision to hospitalize the patient include the known presence of certain etiologic microorganisms (e.g., S. aureus) that are associated with a poor prognosis, multilobe pulmonary involvement, suppurative complications (e.g., empyema or septic arthritis), evidence of poor functional status (e.g., hypotension or hypoxemia on presentation in patients otherwise in classes I, II, and III), evidence of a patient's inability to comply with treatment recommendations, anticipated difficulty in assessing the response to outpatient treatment, and an inadequate home support system that may compromise outpatient care. Discharge from the hospital should be guided by similar considerations.

Most cases of community-acquired pneumonia in otherwise-healthy adults do not require hospitalization. Although desirable, it is often impractical in the outpatient setting to obtain a chest radiograph and sputum Gram's stain and culture in order to confirm the clinical diagnosis of pneumonia and its microbial etiology before starting antimicrobial therapy. Consequently, the oral antimicrobial treatment administered in the outpatient setting is frequently empirical (Table 255-5). The pathogen in such a situation is likely to be M. pneumoniae, S. pneumoniae, or C. pneumoniae. In older patients with underlying chronic respiratory disease, L. pneumophila, H. influenzae, or M. catarrhalis should also be considered. In patients at risk of aspiration, oral anaerobes may be involved. Few oral antimicrobial drugs have a reliable spectrum encompassing all of these pathogens (Table 255-5). Whatever regimen is chosen, its antimicrobial activity should encompass S. pneumoniae, the most common cause of pneumonia. Increasing resistance among pneumococci to all the available oral antimicrobial agents precludes the designation of any one agent as the clear drug of choice.

Table 255-5: Empirical Oral Antimicrobial Therapy for Outpatient Management of Community-Acquired Pneumonia

Strains of S. pneumoniae for which the minimal inhibitory concentration (MIC) of penicillin (as determined by the broth dilution method) is 0.1 to 1.0 g/mL are considered to have intermediate-level resistance, while strains whose MIC is >1.0 g/mL are considered to have high-level resistance. The current, less time-consuming method to screen for penicillin resistance is the use of a 1-g oxacillin disk in a disk diffusion assay. Penicillin resistance (i.e., an MIC 0.1 g/mL) is indicated by a zone of growth inhibition of 19 mm. Antimicrobial gradient paper strips (the E-test), which yield the exact MIC, are as accurate as the broth dilution technique, can be performed as rapidly as the oxacillin disk diffusion assay, and have replaced the oxacillin disk test in many institutions.

The resistance of S. pneumoniae to penicillin varies greatly with the source of the clinical sample tested (e.g., strains isolated from middle-ear fluid are most often resistant), the age of the patient (e.g., resistance is more frequent among children than among adults), the setting (e.g., resistance is more common in day-care centers), the patient's socioeconomic status (the frequency of resistance is highest in samples from suburban and white patients), and the geographic region in which the specimen was collected. Caution must be exercised in the interpretation of surveys of antimicrobial resistance among pneumococci in the United States, which can be strongly affected by these types of sampling bias. In a national survey of clinical isolates from normally sterile body sites that was conducted in 1997 in various surveillance areas throughout the United States by the Centers for Disease Control and Prevention (CDC), 11% (range, 6 to 19%) of 3110 isolates of S. pneumoniae exhibited intermediate-level resistance to penicillin, and 14% (range, 8 to 26%) displayed high-level resistance. However, in another national survey of the antimicrobial susceptibility of clinical isolates obtained from respiratory tract sites between February and June 1997 at 27 U.S. medical centers (SENTRY surveillance program), 28% of 845 isolates (with a range of 11 to 52% at the various medical centers) displayed intermediate-level penicillin resistance, and an additional 16% (with a range of 0 to 33%) displayed high-level penicillin resistance.

As a consequence of the production of altered penicillin-binding proteins with decreased -lactam affinity, penicillin-resistant S. pneumoniae exhibits at least some degree of cross-resistance to all -lactams, including the extended-spectrum third- and fourth-generation cephalosporins. Since the mechanism of penicillin resistance does not involve -lactamase production, -lactam/-lactamase inhibitor combinations (e.g., amoxicillin/clavulanate) offer no advantage. Indeed, the MICs of penicillin and amoxicillin are nearly identical, but the serum levels after equivalent doses are much higher for amoxicillin than for penicillin, a difference that may reflect a therapeutic advantage of amoxicillin. Among the oral cephalosporins, cefaclor, cefadroxil, and cephalexin have variable activity against penicillin-sensitive strains; cefuroxime and cefpodoxime have activity against penicillin-susceptible strains but variable activity against penicillin-intermediate strains and no activity against highly penicillin-resistant strains.

Resistance to other antimicrobial agents, such as the macrolides (erythromycin, clarithromycin, and azithromycin), clindamycin, tetracycline and doxycycline, and trimethoprim-sulfamethoxazole (TMP-SMZ), is also more common among penicillin-intermediate strains than among penicillin-susceptible strains, and it is most common among highly penicillin-resistant strains. Overall rates of resistance among S. pneumoniae strains are ~14% for the macrolides, 4% for clindamycin, up to 10% for tetracyclines, and 20 to 30% for TMP-SMZ. Rates of resistance to the newer fluoroquinolones levofloxacin, gatifloxacin, moxifloxacin, and sparfloxacin are <4%, regardless of penicillin susceptibility. At best, the older fluoroquinolones (e.g., ciprofloxacin) have borderline activity, as judged by serum levels in relation to MICs of these drugs against the pneumococcus.

Optimally, the choice of antimicrobial drugs for empirical therapy should be guided by local resistance patterns, if known. Options for empirical antimicrobial therapy should be modified in light of continually evolving antimicrobial resistance patterns resulting from the introduction of new resistant clones into the community from other regions or the emergence of resistant mutants under the selective pressure of local patterns of antimicrobial use. The IDSA has published guidelines for the treatment of community-acquired pneumonia. These guidelines emphasize the need for a chest radiograph when pneumonia is suspected and for the establishment of a microbial diagnosis (e.g., by sputum Gram's stain with or without culture) whenever possible. Doxycycline and the newer fluoroquinolones are recommended alternatives for initial empirical oral therapy, especially when penicillin-resistant pneumococci are suspected. The utility of the macrolides and amoxicillin depends on susceptibility of pneumococci in the local community.

The regimen should be modified for patients with particular epidemiologic factors or comorbidities related to specific pathogens\em\e.g., structural lung disease or suspected aspiration. Aspiration pneumonia can be treated with amoxicillin/clavulanate, clindamycin, or amoxicillin plus metronidazole because these regimens are active against oral anaerobes. Metronidazole alone has inadequate activity against microaerophilic gram-positive cocci and must be supplemented with a -lactam agent that compensates for this defect in spectrum. If macrolides are used and H. influenzae is suspected, azithromycin or clarithromycin is preferred because of erythromycin's poor activity against this organism. Alternative agents for H. influenzae include amoxicillin/clavulanate, doxycycline, or a fluoroquinolone. The -lactams are not active against pathogens causing atypical pneumonia (e.g., Mycoplasma, C. pneumoniae, or Legionella), in which case doxycycline, a macrolide, or a fluoroquinolone is preferred.

The IDSA guidelines recommend that pneumococcal pneumonia be treated for 7 to 10 days or until the patient has been afebrile for 72 h. Pneumonia caused by Legionella, C. pneumoniae, or Mycoplasma should be treated for 2 to 3 weeks unless azithromycin is used, in which case a 5-day course is acceptable because of the drug's prolonged half-life in tissues.

Patients who have community-acquired pneumonia and are ill enough to be hospitalized (Fig. 255-1) must have a chest radiograph to establish the diagnosis of pneumonia, must undergo prompt microbiologic evaluation (including Gram's staining and culture of sputum and culture of two blood samples drawn by separate venipuncture), and must receive empirical antimicrobial therapy based on Gram's staining of sputum and knowledge of the current antimicrobial sensitivities of the pulmonary pathogens in the local geographic area (Tables 255-6 and 255-7). Antimicrobial therapy should be initiated promptly (e.g., within 8 h of admission). Parenteral antimicrobial therapy in the hospitalized patient is usually mandatory. A lack of sputum production, an atypical clinical presentation, the presence of diffuse radiographic infiltrates, a rapidly progressive downhill course, and a poor response to prior empirical therapy are among the indications for the use of invasive procedures to detect the pulmonary pathogen, especially in the immunocompromised patient. Although broad-spectrum antibacterial therapy should be started during a full evaluation in severely ill patients with rapidly progressing illness, these empirical regimens cannot encompass all the possible pathogens without producing unnecessary toxicity and expense. Indeed, in immunocompromised patients (including those with neutropenia or HIV infection), the number of microbial and noninfectious causes of pulmonary disease is large and increasing. Since failure to provide specific treatment can prove rapidly fatal, a diagnosis should be sought aggressively so that optimal therapy can be started promptly.

Table 255-7: Dosage of Antimicrobial Agents for the Treatment of Pneumonia in Hospitalized Patientsa

Drug Dosage

Ampicillin/sulbactam 3 g IV q6h

Aztreonam 2 g IV q8h

Cefazolin 1-2 g IV q8h

Cefepime 2 g IV q8h

Cefotaxime, ceftizoxime 1-2 g IV q8-12h

Ceftazidime 2 g IV q8h

Ceftriaxone 1-2 g IV q12h

Cefuroxime 750 mg IV q8h

Ciprofloxacin 400 mg IV or 750 mg PO q12h

Clindamycin 600-900 mg IV q8h

Erythromycin 0.5-1.0 g IV q6h

Gentamicin (or tobramycin) 5 mg/kg/d in 3 equally divided doses IV q8h

Imipenem 500 mg IV q6h

Levofloxacin 500 mg IV or PO q24h

Metronidazole 500 mg IV or PO q6h

Nafcillin 2 g IV q4h

Penicillin G 3 million units IV q4-6h

Piperacillin/tazobactam 4.5 g IV q6h

Ticarcillin/clavulanate 3.1 g IV q4h

Vancomycin 1 g (15 mg/kg) IV q12h

*Dosage must be modified for patients with renal failure. Guidelines on the duration of therapy for each pathogen are given in the text of this chapter and of chapters on specific infecting agents.

Penicillin or ampicillin remains the drug of choice for infection due to penicillin-susceptible pneumococci. Studies suggest that high-dose intravenous penicillin G (e.g., 10 to 20 million units daily), ampicillin (2 g every 6 h), ceftriaxone (1 or 2 g every 24 h), or cefotaxime (1 to 2 g every 6 h) constitutes adequate therapy for pneumonia due to strains exhibiting intermediate resistance to penicillin (MIC, 0.1 to 1 g/mL). The effectiveness of high-dose intravenous penicillin against pneumonia due to highly resistant pneumococcal strains is unknown, but MICs of cefotaxime and ceftriaxone for these strains are usually lower than those of penicillin or ampicillin and most other -lactam antibiotics. Ceftriaxone or cefotaxime may be effective when the MIC of penicillin is 1 g/mL and those of ceftriaxone and cefotaxime are 2 g/mL. However, highly cephalosporin-resistant strains have become a problem in certain geographic areas. Since all penicillin-resistant strains are sensitive to vancomycin, initial empirical therapy should include this antibiotic (1 g intravenously every 12 h) when the patient with pneumococcal pneumonia is severely ill, has significant comorbidity, and lives in a region where highly penicillin- or cephalosporin-resistant strains have become common.

If the result of Gram's staining of sputum is not interpretable or not available, then the IDSA guidelines recommend empirical therapy for patients hospitalized on a general medical unit with a -lactam (e.g., ceftriaxone, cefotaxime) or a -lactam/-lactamase inhibitor combination, with or without a macrolide, or with one of the fluoroquinolones alone. Seriously ill patients who are hospitalized in the ICU should always receive a macrolide or a newer fluoroquinolone in addition to the -lactam to cover Legionella. The therapeutic regimens should be modified further in the following situations: structural disease of the lung (e.g., bronchiectasis) requires treatment with an anti-Pseudomonas -lactam plus a macrolide or with a newer fluoroquinolone plus an aminoglycoside; penicillin allergy requires treatment with a newer fluoroquinolone, with or without clindamycin; and suspected aspiration requires treatment with a newer fluoroquinolone plus either clindamycin or metronidazole or with a -lactam/-lactamase inhibitor combination alone. A recent study of almost 13,000 elderly hospitalized patients with pneumonia, which controlled for severity of illness, baseline differences in patient characteristics, and processes of care, documented 30-day mortality that was 26 to 36% lower among those treated initially with a fluoroquinolone alone or a macrolide combined with a second- or nonpseudomonal third-generation cephalosporin than among those initially given a nonpseudomonal third-generation cephalosporin alone. This result may reflect the importance of pathogens such as Mycoplasma, Legionella, and C. pneumoniae in these patients.

Therapy can be switched from intravenous to oral agents within 3 days to complete a 7- to 10-day course if the patient's clinical condition improves rapidly and if antimicrobial agents that are readily absorbed after oral administration and that reach tissue levels above the MIC are available. The presence of S. aureus or aerobic gram-negative bacilli or the development of suppurative complications requires a more prolonged course of therapy. Pneumonia caused by Legionella, C. pneumoniae, or Mycoplasma should be treated for 2 to 3 weeks unless azithromycin is used. Anaerobic lung abscess should be treated with the regimens suggested for aspiration pneumonia until a chest radiograph (with radiography performed at 2-week intervals) is clear or shows only a small stable scar. Therapy is prolonged for 6 weeks to prevent relapse, although shorter courses are probably sufficient for many patients. Surgery is rarely required for lung abscess; indications for surgery include massive hemoptysis and suspected neoplasm. Supportive measures include the administration of supplemental oxygen and intravenous fluids, assistance in clearing secretions, fiberoptic bronchoscopy, and (if necessary) ventilatory support. Caution should be exercised in bronchoscopic drainage of large, fluid-filled lung abscesses because of the potential for sudden massive spillage of large collections of pus into the airways.

Patients with risk factors for HIV infection and an atypical pneumonia syndrome should be evaluated for PCP because of its frequency as an index diagnosis in HIV infection and its potential severity. Tuberculosis and other causes of atypical pneumonia must be excluded as part of the evaluation of these patients. Empirical therapy can consist of either TMP-SMZ (15 to 20 mg of trimethoprim per kg, given daily in four divided doses intravenously or by mouth) or pentamidine (3 to 4 mg/kg daily, given intravenously), and therapy is continued for 3 weeks in confirmed cases of PCP. Although some data suggest that TMP-SMZ is more effective than pentamidine, further studies directly comparing the two agents are needed. The frequency and severity of the adverse effects of the two drugs are generally thought to be equivalent. The addition of glucocorticoids (prednisone, 40 mg twice daily, with subsequent tapering of the dose) early in the course of PCP in patients with an arterial PO2 of <70 mmHg decreases the need for mechanical ventilation and improves the patient's chances of survival and functional status. Prophylaxis for recurrent PCP must be started at the end of therapy.

Pneumonia acquired in institutions such as nursing homes or hospitals is frequently caused by enteric aerobic gram-negative bacilli, P. aeruginosa, or S. aureus, with or without oral anaerobes. Again, the selection of empirical antimicrobial therapy should be guided by Gram's staining of sputum (Tables 255-7 and 255-8) and knowledge of the prevalent nosocomial pathogens and their current in vitro antimicrobial sensitivity patterns in the institution involved. An aggressive diagnostic approach is needed in some circumstances, especially for the immunocompromised patient (as outlined above).

Etiology Regimen (table 255-8)

Presumptive Staphylococcus aureus - Nafcillin or vancomycin

Presumptive enteric aerobic gram-negative bacilli or Pseudomonas aeruginosa

1. Ceftazidime or cefepime ± aminoglycoside
2. Ticarcillin/clavulanate or piperacillin/tazobactam ± aminoglycoside
3. Aztreonam ± aminoglycoside
4. Imipenem± aminoglycoside
5. Fluoroquinolone ± aminoglycoside or b -lactam

Mixed flora

1. Ceftazidime or cefepime + clindamycin (or metronidazole) ± aminoglycoside
2. Ticarcillin/clavulanate or piperacillin/tazobactam ± aminoglycoside
3. Aztreonam + clindamycin (or metronidazole) ± aminoglycoside
4. Imipenem± aminoglycoside5. Fluoroquinolone+ clindamycin (or metronidazole) ± aminoglycoside

S. aureus acquired in some institutions is frequently methicillin resistant. Such strains are resistant to all -lactam antibiotics and may also be resistant to clindamycin, erythromycin, and the fluoroquinolones. Only vancomycin is predictably active against these organisms, and this drug should be added to the empirical regimen when methicillin-resistant organisms may be involved in pneumonia.

When multiantibiotic resistance is a problem, pneumonia due to gram-negative bacilli in the institutionalized patient can be treated initially with a -lactam active against P. aeruginosa (ceftazidime, cefepime, piperacillin/tazobactam, ticarcillin/clavulanate, aztreonam, or imipenem) or with a parenterally administered fluoroquinolone (ciprofloxacin, ofloxacin, gatifloxacin, or levofloxacin). Among the fluoroquinolones, ciprofloxacin remains the most potent antipseudomonal agent. Ticarcillin/clavulanate and piperacillin/tazobactam are preferred over other penicillins with activity against P. aeruginosa (e.g., ticarcillin or piperacillin alone), which are not sufficiently active against Klebsiella pneumoniae, a relatively common pathogen. However, for infection suspected to be due to P. aeruginosa, the higher dose recommended by the package insert is required; a lower dose contains less piperacillin or ticarcillin than is needed to be effective against this organism. Ampicillin/sulbactam, the other parenterally administered -lactam/-lactamase inhibitor combination, is not active against many nosocomial pathogens, such as P. aeruginosa, Enterobacter spp., and Serratia spp., and therefore is inappropriate as empirical therapy for nosocomial pneumonia.

In seriously ill patients, especially those infected with organisms in which resistance frequently emerges during therapy (e.g., P. aeruginosa), use of a -lactam/aminoglycoside or -lactam/fluoroquinolone combination is prudent. Combinations of a -lactam plus an aminoglycoside are used for bactericidal synergy. Combinations of a -lactam or an aminoglycoside with a fluoroquinolone are not expected to enhance the already-rapid bactericidal activity of the fluoroquinolone alone. However, such combinations are also used to broaden the spectrum of antibacterial activity, to cover the possibility of infection with resistant pathogens, to treat polymicrobial infection, and to prevent the emergence of antimicrobial resistance.

Pneumonia due to possible coinfection with aerobic gram-negative bacilli and anaerobes, as reflected by a polymicrobial flora on Gram's staining of sputum, can usually be treated with any of the following regimens: (1) cefepime or ceftazidime plus metronidazole or clindamycin, (2) aztreonam or a fluoroquinolone plus clindamycin, or (3) imipenem, piperacillin/tazobactam, or ticarcillin/clavulanate. The regimens should include double coverage for P. aeruginosa when this organism is suspected (Table 255-8).

The production of chromosomally encoded, inducible -lactamases by some aerobic gram-negative bacilli, including Serratia marcescens, Enterobacter cloacae, Citrobacter freundii, Morganella morganii, P. aeruginosa, and Acinetobacter calcoaceticus, has important implications for the treatment of nosocomial pneumonia in institutions where these organisms are common nosocomial pathogens. Antibiotic resistance in these pathogens has been attributed to two related mechanisms: inducible production of chromosomally encoded -lactamases and selection of mutants that have lost the genes that control expression of -lactamase production. The control genes repress -lactamase production in the absence of a -lactam agent and allow -lactamase production in the presence of a -lactam agent. This group of organisms has a relatively high mutation rate for loss of these control genes, and their loss results in continuous production of large amounts of -lactamase (stable derepression). The derepressed mutants are resistant to third-generation cephalosporins, aztreonam, and broad-spectrum penicillins. These chromosomally encoded, inducible -lactamases are not inhibited by clavulanic acid, tazobactam, or sulbactam.

Selection by the -lactam antibiotic of the derepressed mutants present in the dense bacterial populations of infected pulmonary tissue at the initiation of antibiotic therapy apparently accounts for the emergence of resistance during therapy, which is especially problematic in severely compromised patients whose defective host defenses are unable to control the growth of a few resistant mutants. The only -lactam agents that maintain activity against the derepressed mutants are the fourth-generation cephalosporin cefepime and the carbapenem imipenem. The fluoroquinolones and aminoglycosides may also retain activity against these mutants. TMP-SMZ may remain active against all of these gram-negative bacilli except P. aeruginosa, which is inherently resistant to this agent. Some clinicians have questioned the efficacy of aminoglycosides alone for the treatment of gram-negative bacillary pneumonia. The poor clinical efficacy of aminoglycosides has been attributed to the low drug levels attained in bronchial secretions and to a loss of antimicrobial activity due to the relative acidity of purulent secretions, the anaerobic conditions in infected lung, and (in the case of P. aeruginosa) the divalent cations calcium and magnesium. The nephrotoxicity and ototoxicity of aminoglycosides frequently lead to underdosing with these agents. These problems are compounded by unpredictable pharmacokinetics that necessitate measurement of serum levels of aminoglycosides. If multiantibiotic-resistant nosocomial organisms are likely to be the pathogens infecting severely compromised patients, reliable empirical agents may be fluoroquinolones, cefepime, and imipenem-unless resistance to these drugs is also endemic in the institution. Some strains of K. pneumoniae and Escherichia coli have acquired a plasmid encoding the production of an extended-spectrum -lactamase that can be detected as in vitro resistance to ceftazidime or aztreonam. The presence of an extended-spectrum -lactamase confers resistance to all third-generation cephalosporins and aztreonam. Some of these strains may also be resistant to piperacillin/tazobactam and cefepime, and many are also resistant to the fluoroquinolones. The only reliable agents are the carbapenems, such as imipenem. Up-to-date knowledge of the antimicrobial sensitivities of an institution's nosocomial pathogens and use of various preventive practices are mandatory.

Amantadine (200 mg/d for most adults and 100 mg/d for persons >65 years of age) is effective for the prevention of influenza A virus infection in the unimmunized patient during an influenza A outbreak and for the treatment (for 5 to 7 days) of early influenza A virus infection. Ribavirin is effective for respiratory syncytial virus infection. Intravenous acyclovir (5 to 10 mg/kg every 8 h for 7 to 14 days) is appropriate for varicella pneumonia. Treatment of cytomegalovirus pneumonia has yielded unsatisfactory results, but intravenous immunoglobulin combined with ganciclovir may be effective in some instances. Therapy for hantavirus pulmonary syndrome is supportive, and overall mortality has been 55%.

The prevention of pneumonia involves either (1) decreasing the likelihood of encountering the pathogen or (2) strengthening the host's response once the pathogen is encountered. The first approach can include measures such as hand washing and glove use by persons who care for patients infected with contact-transmitted pathogens (e.g., aerobic gram-negative bacilli); use of face masks or negative-pressure isolation rooms for patients with pneumonia due to pathogens spread by the aerosol route (e.g., M. tuberculosis); prompt institution of effective chemotherapy for patients with contagious illnesses; and correction of conditions that facilitate aspiration. The second approach includes the use of chemoprophylaxis or immunization for patients at risk. Chemoprophylaxis may be administered to patients who have encountered or are likely to encounter the pathogen before they become symptomatic (e.g., amantadine during a community outbreak of influenza A, as mentioned above; isoniazid for tuberculosis; or TMP-SMZ for pneumocystosis) or to patients who are likely to have a recurrence following recovery from a symptomatic episode (e.g., TMP-SMZ for pneumocystosis in patients with HIV infection). The prevention of nosocomial pneumonia requires good infection control practices, judicious use of broad-spectrum antimicrobial agents, and maintenance of patients' gastric acidity-a major factor that prevents colonization of the gastrointestinal tract by nosocomial gram-negative bacillary pathogens. To prevent stress ulceration, it is preferable to use sucralfate, which maintains gastric acidity, rather than H₂-blocking agents. To prevent ventilator-associated nosocomial pneumonia, the following strategies have been proposed: use of the semirecumbent position, of endotracheal tubes that allow continuous aspiration of secretions accumulating above the cuff, and of heat and moisture exchangers that reduce the formation of condensate within the tubing circuitry. Vaccines (Chaps. 122, 138, 149 and 190194) are available for immunization against S. pneumoniae, H. influenzae type b, influenza viruses A and B, and measles virus. Influenza vaccine is strongly recommended for individuals >55 years old and pneumococcal vaccine for those >65 years old; these vaccines should be administered to persons of any age who are at risk of adverse consequences of influenza or pneumonia because of underlying conditions. Pneumococcal, Haemophilus, and influenza vaccines are recommended for HIV-infected patients who are still capable of responding to a vaccine challenge. The currently available 23-valent pneumococcal vaccine covers 88% of the serotypes causing systemic disease as well as 8% of related serotypes. The increasing prevalence of multiantibiotic resistance among pneumococci makes pneumococcal immunization of high-risk individuals of utmost importance. Immune serum globulin is available for intravenous replacement therapy in those patients with congenital or acquired hypogammaglobulinemia. Some patients who have selective IgG2 subtype deficiency and recurrent sinopulmonary infections and who are immunologically unresponsive to capsular polysaccharide vaccines may nevertheless have an antibody response to the capsular polysaccharide that is covalently linked to a protein, as it is in the conjugate H. influenzae type b vaccine and a similar experimental conjugate pneumococcal vaccine.

BACK