Chapter 51. Pneumonia

• Pneumonia is one of the leading causes of respiratory failure leading to admission to the intensive care unit and is by far the most common nosocomial infection in critically ill patients.
• Mortality rate may exceed 50% for severe community-acquired pneumonia (CAP) requiring admission to the intensive care unit and varies from 30% to 60% in nosocomial cases.
• The initial approach to diagnostic testing and empiric therapy is guided by the clinical presentation of the pneumonia, which can be categorized as follows: CAP, acute CAP, aspiration pneumonia/lung abscess, and chronic pneumonia; nosocomial pneumonia; and pulmonary infiltrate in immunocompromised host.
• Investigation of acute pneumonia should always include a blood culture and Gram stain and culture of lower respiratory tract secretions. Smear and culture for Legionella species and serologic testing for other atypical pathogens should be done when atypical pneumonia features are present and in otherwise undiagnosed cases of severe pneumonia.
• Fiberoptic bronchoscopy with quantitative bacteriology of a protected brush specimen or bronchoalveolar lavage specimen is the procedure of choice for diagnosis of acute CAP and nosocomial pneumonia that respond poorly to treatment and resist diagnosis by noninvasive means.
• Empiric intravenous antimicrobial therapy should be begun immediately in all critically ill patients with an acute pneumonia. Antimicrobial coverage should include all pathogens of more than trivial probability and should always include Streptococcus pneumoniae, Staphylococcus aureus, and Enterobacteriaceae. In cases of severe pneumonia of unknown cause, a fluoroquinolone should be included to cover Legionella species, and dual antimicrobial coverage for Pseudomonas species should be given to patients at special risk of carrying this organism.
• Empiric antimicrobial therapy initially can be withheld in less severely ill patients in whom the diagnosis of infectious pneumonia is in doubt and in patients with a chronic pneumonia presentation, pending definitive diagnosis.
• Pulmonary infiltrates in an immunocompromised host may be caused by any of the infectious agents affecting the normal population, by opportunistic infections, or by noninfectious processes. The required approach combines empiric therapy, selected on the basis of the particular infectious agents for which the patient is at special risk, with a timely stepwise diagnostic approach that moves from noninvasive testing to fiberoptic bronchoscopy to open lung biopsy at a rate determined by the rate of progression of the patient's undiagnosed pneumonia.

Pneumonia remains the leading infectious cause of death in the developed world. It is also the most frequent infection leading to admission to most intensive care units (ICUs) and by far the most important nosocomial infection complicating the treatment of patients admitted to ICUs for other problems. Roughly 5% to 10% of patients hospitalized for community-acquired pneumonia (CAP) require ICU admission, with a mortality rate of 30% to 50%.1–3 Pneumonia is also the most important nosocomial infection leading to ICU admission and a frequent complication of treatment in an ICU, occurring in 5% to 50% of patients requiring mechanical ventilation.4 A systematic approach and an aggressive attitude toward diagnosis and treatment of pneumonia therefore are fundamental to good critical care practice.

The discussion that follows categorizes the major pneumonia syndromes as CAP, nosocomial pneumonia, and pulmonary infiltrates in the immunocompromised host. Whereas the etiologies of pneumonia in these three categories overlap to some degree, the underlying pathogeneses, etiologic differential diagnoses, and approaches to empiric antimicrobial therapy differ considerably.

Pneumonia in patients presenting from outside the hospital can lead to ICU admission for a variety of reasons: hypoxemic or hypercapnic respiratory failure, depressed level of consciousness caused by hypoxia or sepsis, or hypotension related to relative hypovolemia or septic shock. Although supportive management for these problems is generally similar for all types of pneumonia and the principles of circulatory and respiratory management differ little from those for shock and respiratory failure of other causes, there are major differences among the different forms of pneumonia in the approach to diagnosis and empiric antimicrobial therapy. The great majority of CAPs are caused by common bacterial pathogens such as Streptococcus pneumoniae, Staphylococcus aureus, and Haemophilus influenzae. However, in most series of pneumonia severe enough to require ICU admission, a significant minority of cases is caused by a number of less common pathogens with widely differing antimicrobial susceptibilities.1–7 Approximate percentages for the major causes of CAP are shown in Table 51-1. The distribution of etiologies varies widely among different geographic areas.

Pathogenesis

Pathogenic mechanisms of pneumonia vary greatly among the different etiologic agents, depending on the relative virulence of the organism, which host defenses are most important in preventing infection with that organism, and the common route of entry of the organism into the lung. The three major routes of entry are aspiration of oropharyngeal or gastric contents into the lung, inhalation of aerosols or particles containing organisms, and hematogenous spread of organisms into the lung from another infected site.

Aspiration accounts for the vast majority of pulmonary infections, with the particular clinical syndrome being determined by the quantity of material aspirated, the nature of the bacteria in the aspirated material, and the efficiency of the host's pulmonary defense mechanisms. The upper airway of normal human beings is frequently colonized by S. pneumoniae, H. influenzae, and, occasionally, S. aureus, in addition to the normal microflora consisting of mixed aerobic and anaerobic bacteria. Colonization by enteric gram-negative bacilli, such as Escherichia coli or Klebsiella pneumoniae, is much less frequent, occurring in fewer than 1% of persons in the community at large, although it is more frequent in alcoholics, persons with poor oral hygiene, and the institutionalized elderly.8,9 Even people who have normal upper airway reflexes aspirate small quantities of oropharyngeal material during sleep.10 Normally, the organisms so aspirated are cleared by the bronchial mucociliary clearance mechanism, and any residual debris is cleared by the phagocytic cells present in alveoli and on the bronchial mucosa. However, if the aspirate contains organisms with significant potential for virulence, and if the aspiration occurs in a human host with impaired pulmonary clearance mechanisms, the organism can proliferate in the pulmonary parenchyma and cause pneumonia. Pneumonia caused by relatively high-grade pathogens such as S. pneumoniae, S. aureus, and enteric gram-negative bacilli generally occurs as an acute infection without clinically obvious major aspiration (i.e., the aspiration is subclinical) in patients with damaged or deficient host defenses. Examples include pneumonia after a viral respiratory infection (which impairs ciliary clearance mechanisms), pneumonia in alcoholics or others with intermittently depressed level of consciousness (greater opportunity for subclinical aspiration), and pneumonia in patients with impaired mucociliary clearance (chronic obstructive pulmonary disease, or chronic bronchitis) or relative immunologic impairment (diabetes mellitus, uremia, or hypogammaglobulinemia).

The normal microflora of the upper airway has limited virulence and therefore causes pulmonary infection mainly when aspiration is more florid, when the aspirated oropharyngeal material is present in greater quantity, or when there is severe local impairment of tracheobronchial clearance. Examples of these conditions include risk factors for major aspiration, such as uncontrolled epileptic seizures or abnormal motor control of the upper airway; periodontal disease; and bronchial obstruction caused by foreign body aspiration or neoplasm. The infection produced by these relatively nonpathogenic bacteria is usually an indolent one producing a subacute, but necrotizing, pneumonia, with eventual liquefaction of pulmonary parenchyma and abscess formation.

Because the numbers of microorganisms reaching the lung usually are extremely small when the route of entry is inhalation, only organisms that are highly efficient pathogens produce infection by this mechanism. This does not include most of the common bacterial pathogens but does include most respiratory viruses, Legionella species, Mycoplasma pneumoniae, Chlamydia species, Coxiella burnetii, Mycobacterium tuberculosis, and most of the endemic fungal pneumonias. These pathogens are inhaled and deposited within alveoli, evading tracheobronchial mucociliary clearance. For the most part they share the ability to resist phagocytosis or to survive intracellularly within phagocytes, requiring the development of a humoral and cell-mediated immune response to limit infection.

For hematogenous spread of infection to the lung, the organisms must gain access to the venous circulation; the organisms lodge in the pulmonary microvasculature of the lung and proliferate. The pneumonia is usually diffuse or multinodular throughout both lung fields. In most cases, there is an established infection elsewhere, which is then carried to the lung, or organisms are injected directly into the bloodstream, as with intravenous drug abusers and persons given contaminated intravenous fluid infusions. The common pathogens are S. aureus and aerobic gram-negative bacilli such as Pseudomonas aeruginosa. A few rare pulmonary infections caused by very virulent organisms that gain entry directly through damaged skin are also transmitted to the lung hematogenously; such pulmonary involvement may occur in tularemia, brucellosis, and melioidosis.

Pathophysiology

The proliferation of microorganisms within the pulmonary parenchyma elicits the host's full acute inflammatory response, with exudation of protein-rich fluid and influx of large numbers of phagocytic cells into alveoli and airways. The local mechanical consequences of this include impaired distribution of ventilation and a decrease in lung compliance, which contribute to increased work of breathing and the symptom of dyspnea. Ventilation/perfusion mismatch is increased somewhat, particularly in predominantly interstitial pneumonias; in most acute bacterial pneumonias, the major mechanism for arterial hypoxemia is intrapulmonary shunt caused by maintenance of pulmonary arterial blood flow to consolidated lung.11,12 There is also evidence that metabolically active inflammatory cells within the consolidated lung consume oxygen, thus further decreasing pulmonary venous oxygen content and arterial oxygenation.12–14

The local inflammatory response includes activation of mononuclear phagocytic cells, which are the main source of interleukin 1, cachectin, and other cytokines that act as hormones mediating the acute-phase response. These effects result in fever, leukocytosis, and the many other metabolic changes of infection. In more severe cases, bacteremia or microbial antigenemia may activate the inflammatory response systemically, leading to frank septic shock (see Chap. 46).

Clinical and Radiographic Features

The basic features common to most forms of pneumonia include (a) the presence of a systemic inflammatory response manifest as fever, elevation of the white blood cell count, and, in severe cases, other features of the sepsis syndrome; (b) pulmonary symptoms such as cough, sputum production, or hemoptysis, dyspnea, and pleuritic chest pain (if the pleura is involved); (c) physical findings consistent with an inflammatory pulmonary parenchymal process, such as tachypnea, rales, or signs of consolidation (bronchial breathing, dullness to percussion, increased vocal fremitus over the consolidated lung region); (d) evidence of abnormal lung function, such as arterial hypoxemia and hypocapnia (in a hyperventilating, dyspneic patient) or hypercapnia (in a patient with acute-on-chronic respiratory failure); and (e) a radiographic pulmonary infiltrate consistent with pneumonia.

There are major differences in the time course, frequency, and pattern of these findings among the many forms of pneumonia caused by the multiplicity of etiologic agents. Characterizing the patient's presentation based on the clinical, epidemiologic, and radiographic features—information generally available for most patients when they are initially seen—is a useful starting point in the assessment of pneumonia. These clinical presentations include:

1. Acute CAP

2. Aspiration pneumonia

3. Chronic pneumonia

Acute Community-Acquired Pneumonia

Bacterial Pneumonia

The typical history for CAP caused by conventional bacteria, most commonly S. pneumoniae, is acute onset of fever, with chills or rigors, associated with dyspnea and a cough productive of purulent or bloody sputum, sometimes with pleuritic chest pain. Patients with no chronic underlying disease often report a preceding upper respiratory tract infection, but most pneumonias occur in the presence of a predisposing illness, such as alcohol abuse, chronic lung disease, chronic cardiac or renal failure, or malignancy. In more severely debilitated patients or the institutionalized elderly, a specific history may be unavailable; new or worsened confusion and tachypnea noted by attendants is often the only history in these instances.

Physical examination shows a tachypneic, apprehensive patient with tachycardia, fever, and diaphoresis. Crackles are heard over the involved lung fields on auscultation. There may be signs of pulmonary consolidation (bronchial breathing, dullness to percussion, or increased tactile fremitus).

The chest radiograph usually demonstrates airspace consolidation with a lobar or a bronchopneumonic pattern. A dense lobar infiltrate, particularly with an air bronchogram, very strongly suggests acute bacterial pneumonia (Fig. 51-1); bronchopneumonic and patchy infiltrates are less specific. Interstitial infiltrates virtually exclude the usual bacterial pneumonias, although diffuse reticulonodular patterns or multicentric pneumonia sometimes occur with hematogenous pneumonia (e.g., in intravenous drug abusers; Fig. 51-2).

The polymorphonuclear leukocyte count is usually elevated and shifted leftward but may be depressed in severely ill patients with shock or may even be within the normal range in debilitated or elderly patients.

Legionnaire's Disease

Whereas early descriptions of community-acquired Legionnaire's disease emphasized many clinical features that set it apart from usual bacterial pneumonia,15 we now know that Legionnaire's disease can present in ways clinically indistinguishable from disease caused by pneumococcus.16 Clinical features that may suggest the diagnosis include dry cough, preceding diarrhea or other gastrointestinal (GI) symptoms, and encephalopathy not explained by other features of the illness. Epidemiologic clues include exposure to aerosols of Legionella-contaminated water in cooling towers or potable water distribution systems, underlying chronic lung disease, and immunologic impairment, particularly that caused by corticosteroid therapy. In patients requiring intensive care, the course is usually one of rapidly progressive pneumonia, sometimes with the aforementioned clinical features, or extrapulmonary involvement complicated by acute hypoxemic respiratory failure. The radiologic abnormalities are variable; most characteristic is peripheral rounded airspace consolidation with ill-defined margins (Fig. 51-3); however, patchy infiltration is also common.17

Mycoplasma pneumoniae Pneumonia

The term atypical pneumonia was coined to describe the pneumonia syndrome caused by M. pneumoniae, a pneumonia usually occurring in people in the second and third decades of life without major underlying disease predisposing to acute bacterial pneumonia. The clinical syndrome consists of a prodromal upper respiratory tract infection followed by an increasingly persistent dry cough associated with low-grade fever, frequently with extrapulmonary symptoms such as diarrhea, myalgia, arthralgia, and skin rash. In severe cases, the patient is tachypneic and cyanotic, but examination of the chest generally shows only widespread crackles without signs of pulmonary consolidation. The pharynx may be red; bullous myringitis is occasionally present, and a rash is sometimes seen. The chest radiograph most typically shows a unilateral segmental infiltrate but may show patchy or interstitial infiltrates bilaterally. In patients requiring intensive care for respiratory failure caused by this infection, diffuse bilateral interstitial and alveolar infiltrates are usual (Fig. 51-4). Other extrapulmonary features can also lead to the need for intensive care, in particular encephalomyelitis or myocarditis (manifest more commonly as prolonged Qt interval or ventricular arrhythmias than as congestive heart failure). A moderately “septic” appearance and hemodynamic assessment is usual; however, frank septic shock is not, and hypotension is more often caused by volume depletion.

In the more severe cases seen in the ICU, the polymorphonuclear leukocyte count is elevated and shifted leftward, and thrombocytopenia may be present. Hemolytic anemia may be present because of the presence of high-titer cold agglutinins, which may also produce erroneous red cell indices, or “error flags,” on automated cell counters because of red cell agglutination. Moderate elevation of hepatic transaminase levels is common. Although most patients do not produce sputum, in those who do it is usually purulent grossly and on microscopic examination; organisms are conspicuously absent on Gram stain.

Chlamydia psittaci

The major epidemiologic clue to psittacosis is exposure to infected birds, often pets (budgerigars and other related species which may or may not be visibly ill) in homes or pet stores, or commercial birds, such as turkeys, which may shed the organism when killed and eviscerated in processing plants.18 The pneumonia follows exposure by 1 to 2 weeks. High fever and a persistent dry cough are frequently associated with a variety of atypical features: myalgias, headache, GI symptoms, and occasionally a macular rash. On auscultation of the lungs, crackles over involved lung regions, without signs of pulmonary consolidation, are usual. Extrapulmonary physical findings are not infrequent: hepatomegaly, splenomegaly, pleural and pericardial friction rubs, and, in the rare case with complicating endocarditis, pathologic cardiac murmurs. In severe cases, dyspnea and hypoxia are prominent; encephalopathy, with confusion, obtundation, or even coma, may occur. The pulmonary infiltrate is generally a patchy infiltrate in all lung fields, with lower lobe predominance, but this is too variable to be a major differential point. Other routine laboratory studies are equally nonspecific, demonstrating only the usual hematologic and biochemical changes common in patients with serious infections.

Other Chlamydia species may also cause pneumonia (C. pneumoniae commonly, C. trachomatis rarely) but generally do not cause acute CAP sufficiently severe to require intensive care, unless the disease occurs in patients with severe underlying debility.

Q Fever Pneumonia

Coxiella burnetii is a rickettsia-like organism that is transmitted to human beings by inhalation of aerosols or suspended particulate matter from a wide variety of domestic and wild animals, including sheep, goats, cattle, domestic fowl, mice, rabbits, and parturient cats.19 Usually the infected animal is not ill. The illness begins with fever and chills, associated with myalgia and severe diffuse headache in most cases. Some persons with Q fever have GI symptoms. Cough is usually not severe and is nonproductive; however, in cases likely to be referred for intensive care, dyspnea and hypoxemia are prominent. Peripheral segmental infiltrates or rounded opacities are the most common radiographic findings, but no radiologic pattern excludes the diagnosis. Leukocytosis, elevated hepatic transaminase levels, and electrolyte disturbances are common but not specific.

Viral Pneumonia

In adults, the major viral causes of pneumonia that produce illnesses severe enough to lead to ICU admission are influenza viruses A and B. Rarely, severe pneumonia can be caused by respiratory syncytial virus (RSV) or cytomegalovirus (CMV), especially in the debilitated elderly or the immunocompromised; by Varicella zoster virus in healthy adults with primary chicken pox or secondary to disseminated herpes zoster in the immunocompromised; by Epstein-Barr virus in nonimmunocompromised adults with unusually severe infectious mononucleosis; or by adenovirus in susceptible young adults in close contact (e.g., military recruits). Severe influenza occurs during epidemic periods of the year (“flu season,” usually winter or early spring), particularly in nonimmunized patients who are elderly or have significant underlying disease. A “flu-like” illness, with fever, myalgia, headache, sore throat, and harsh cough with burning retrosternal chest pain, is followed by increasing dyspnea and prostration. Physical findings include pharyngitis, crackles and wheezes in all lung fields, tachypnea, and cyanosis. The chest radiograph generally shows diffuse interstitial infiltrates (Fig. 51-5) unless there is also a complicating acute bacterial pneumonia, which is a common circumstance.20

In autumn 2002, an apparently novel, highly contagious form of severe pneumonia appeared in southern China and over the ensuing months was transmitted by air travelers throughout the world, leading to multiple local outbreaks of different degrees of severity.21 Termed severe acute respiratory syndrome (SARS), the illness has been shown to be caused by a coronavirus of animal origin transmitted to humans.22 Typically, after close contact with an infected individual, there is an incubation period of 2 to 10 days followed by abrupt onset of fever with rigors.23 Three to 5 days after the onset of fever, there follows the development of dyspnea associated with headache, malaise, and dry cough. Physical examination at this time is unremarkable except for crackles and signs of consolidation on chest examination. The white blood cell count is generally unremarkable except for moderate lymphopenia. The chest radiograph may be unremarkable in less severe cases, but in definite cases shows airspace opacities of variable (focal or patchy) distribution. In the more severe cases, pulmonary infiltrates progress over 5 to 10 days to full-blown acute respiratory distress syndrome (ARDS) radiologically and histopathologically, which requires intensive care for mechanical ventilatory support. Mortality rate worldwide has averaged about 15%, ranging from 1% to 5% in younger, previously well, people to higher than 50% in the elderly or debilitated. At the time of this writing, ongoing transmission of this infection appears to be halted; however, the potential for re-emergence of the disease clearly exists, requiring physicians to remain alert to this possibility when faced with rapidly advancing undiagnosed pneumonia in a returned traveler.

Pneumocystis carinii Pneumonia

Although this infection is an opportunistic pneumonia occurring only in the severely immunocompromised, the current prevalence of the human immunodeficiency virus (HIV) among people who previously were well requires the physician to consider it as a possible diagnosis in patients presenting with diffuse pneumonia even in the absence of a prior diagnosis of HIV infection. It presents as a relatively insidious illness characterized by low-grade fever, dry cough, and increasing dyspnea, usually with diffuse infiltrates on the chest radiograph (see Chap. 48). Although Pneumocystis carinii pneumonia (PCP) would be the immediate consideration with this presentation in a patient with known advanced HIV infection, the diagnosis is less obvious when this diagnosis has not been made, particularly when the patient is not known to be at increased risk for HIV infection (high-risk sexual activity, prior transfusion with potentially HIV-contaminated blood products, illicit intravenous drug use, or exposure to multiply-used nonsterile needles, etc.).

Aspiration Pneumonia

Frank aspiration of oropharyngeal or gastric contents can produce any of a number of different clinical presentations, depending on the amount and nature of the material aspirated.24 Large-particle aspiration causes airway obstruction with acute asphyxia (the “cafe coronary”), atelectasis, unilateral or localized hyperinflation, or, subacutely, bacterial pneumonia or lung abscess caused by impaired mucociliary clearance in the airway obstructed by a foreign body. Small-particle or liquid aspiration usually causes a low-grade chemical pneumonitis, with a pulmonary infiltrate that clears over a few days without treatment. When the aspirated liquid is gastric acid, a more severe chemical burn is the result, with rapid exudation of fluid into the lung. Consequences can include acute hypovolemic hypotension and, if the region of damaged lung is relatively large, acute hypoxemic respiratory failure.

In most cases of witnessed pulmonary aspiration, the number and pathogenicity of aspirated bacteria are low, and no infection requiring immediate antimicrobial therapy is present. In addition, no evidence indicates that administering antimicrobials at this point decreases the risk of subsequent infective pneumonia, which, when it occurs, becomes evident 3 days to 1 week after the episode of aspiration.24,25 A major exception to this rule is aspiration of gastric contents that have become heavily contaminated with mixed enteric organisms. This occurs in the setting of bowel obstruction with accumulation of large amounts of feculent upper GI fluid, which may be aspirated. It may also occur after upper abdominal surgery or in the presence of upper GI bleeding or paralytic ileus, particularly when there is absence of the normal low gastric pH on account of achlorhydria or treatment with antacids or histamine-blocking agents. Aspiration of feculent gastric contents can result in a fulminant necrotizing pneumonia caused by mixed aerobic and anaerobic enteric bacteria. Prompt broad-spectrum antimicrobial therapy is mandatory.

A much more common syndrome related to pulmonary aspiration is pneumonia or lung abscess caused by the predominantly anaerobic normal microflora of the mouth. Persons at particular risk are those who have a propensity toward aspiration because of a continuously or intermittently depressed level of consciousness (alcohol or drug abuse, epileptic seizures, stupor or coma of any cause), an impaired swallowing mechanism (esophageal or neurologic disease), or an impairment of tracheobronchial clearance mechanisms (endobronchial lesion or foreign body). After the aspiration episode, a low-grade pneumonia, manifest as cough, fever, and malaise, begins within a few days to 1 week. A chest radiograph at this time usually shows a patchy infiltrate, which may be unilateral or bilateral, occurring a little more commonly in the right lung and predominantly in lung zones that would be dependent in the supine patient (posterior segments of the upper and lower lobes). If the patient does not seek medical attention at this stage, the illness evolves over 1 to 4 weeks, with a persistent fever with night sweats, malaise, anorexia, and weight loss. The onset of production of large amounts of foul-smelling, watery sputum signals the development of a lung abscess, manifest on a chest radiograph as a cavity (or multiple smaller cavities), often with an air–fluid level, surrounded by a pulmonary infiltrate (Fig. 51-6). Involvement of the pleura often results in empyema, as demonstrated by clinical examination and radiographic findings consistent with a pleural effusion related to the region of pulmonary consolidation. Most patients with this syndrome are subacutely, rather than acutely, ill, but very large empyema collections or relatively extensive pneumonia can produce respiratory failure. In particular, the onset of respiratory insufficiency caused by re-expansion pulmonary edema after drainage of large empyemas may require intensive supportive therapy.

Leukocytosis may be mild or marked; many patients are anemic, especially those with a more protracted course. Electrolyte disturbances, in particular hyponatremia, are common. A variety of biochemical abnormalities related to lengthy illness and malnutrition may be present, including hypoalbuminemia, hypophosphatemia, and hypomagnesemia. None of these is particularly valuable in categorizing the syndrome.

Chronic Pneumonia

Pneumonia in intensive care practice is predominantly an acute problem; in a significant minority of cases, however, a chronic progressive pulmonary process presents with impending or established respiratory failure or with circulatory consequences of the inflammatory process. It is critically important to recognize the presentation as subacute or chronic, because the approaches to diagnosis and empiric therapy are entirely different from those employed in patients with acute pneumonia. Because the differential diagnosis can be extremely large, including many relatively exotic infectious diseases and a number of noninfectious entities, a systematic and aggressive diagnostic approach is more rewarding than wide-spectrum empiric therapy, the initial approach generally adopted for acute pneumonias.

The usual definition of chronic pneumonia is progressive pulmonary symptoms (cough, dyspnea, and pain) for a period of 3 weeks to several months, associated with radiographic evidence of an inflammatory pulmonary parenchymal process. The patient generally appears chronically, rather than acutely, ill and is often malnourished or cachectic. Hematologic and routine biochemical testings show the usual evidence of a longstanding inflammatory process: mild leukocytosis, lymphopenia, and anemia, more often with thrombocytosis than with thrombocytopenia; hypoalbuminemia; and multiple mild electrolyte abnormalities. Depending on the cause of the process and the specific other organs involved, a wide variety of other findings may be present.

Diagnosis

Pneumonia is often an obvious diagnosis, particularly when respiratory symptoms associated with a new pulmonary infiltrate and systemic signs of infection occur in a previously well patient. However, in patients with underlying lung or cardiac disease, the diagnosis may be more difficult. Increased dyspnea and cough with low-grade fever in a patient with an underlying chronic lung disease associated with long-term abnormalities on chest radiograph may be caused by pneumonia but may also represent a viral bronchitis or even an infection or inflammatory process elsewhere in the body. In a patient with known congestive cardiac failure, respiratory deterioration with a moderate leukocytosis and asymmetric radiographic findings of pulmonary edema may be due simply to heart failure, but the radiographic asymmetry may be caused by an underlying pneumonia. These less certain situations can be approached in one of two ways, depending on a clinical assessment of the relative likelihood that superimposed pneumonia is present, the severity of the illness, and the potential consequences of a wrong judgment. When pneumonia seems less likely and the patient is relatively stable, it is often reasonable to treat the underlying condition and withhold antimicrobial therapy with the expectation that the patient will improve without it. In the patient with congestive heart failure, steady clearing of the pulmonary edema, including the asymmetric region, without increasing systemic signs of infection helps to exclude the diagnosis of pneumonia; but failure of the local infiltrate to clear with diuresis, persistent productive cough, fever, and leukocytosis mandate further investigation and empiric treatment for pneumonia. In a more severely ill patient in whom it is felt that any delay in treatment for pneumonia would adversely affect outcome, it is more prudent to initiate empiric antimicrobial therapy at the outset and to stop that therapy later if the clinical course does not support the initial impression of possible pneumonia.

Cough with sputum production, fever, and leukocytosis without a convincing pulmonary infiltrate on chest radiograph is usually caused by viral bronchitis or, in the patient with chronic lung disease, an exacerbation of chronic bronchitis caused by a viral infection, or increased endobronchial bacterial microflora. However, in occasional cases, pneumonia may be present but not visible on the radiograph. This is usually caused by intravascular volume depletion delaying the appearance of the infiltrate (Fig. 51-7) or by a technically inadequate radiograph. Repeating the radiograph with a better technique or after volume resuscitation usually clarifies these situations.

When one has made a clinical diagnosis of pneumonia, the next major issues are making an etiologic diagnosis and prescribing antimicrobial therapy. These tasks can be made difficult by the fact that, although most pneumonias are caused by a small number of common pathogens, a significant minority is caused by a wide range of less common pathogens that require different approaches to investigation and treatment. The differential diagnosis also includes a large number of noninfectious causes of pulmonary infiltrates associated with an acute inflammatory response. For acute pneumonias, these include pulmonary atelectasis, chemical pneumonitis from aspiration or toxic inhalation, pulmonary infarction from pulmonary embolism, lung contusion in trauma cases, and a range of immunologically mediated acute pneumonitis syndromes. Noninfectious causes of subacute and chronic pneumonias are even more common. Many of these diagnoses can be suspected on the basis of the clinical context; in severe cases otherwise consistent with an infectious etiology, one usually should proceed with investigation and initial empiric therapy for infection until such an etiology is excluded.

The first step in the diagnostic assessment is to try to categorize the pneumonia presentation as outlined under Clinical and Radiographic Features. The diagnostic and immediate therapeutic approach to each is different: severe acute CAP mandates empiric antimicrobial therapy and investigation directed at identifying the offending organism; aspiration pneumonia or lung abscess is generally treated empirically with agents directed at the predominantly anaerobic, normal oral microflora, with investigation aimed mainly at uncovering an underlying anatomic cause for the pneumonia; and the chronic pneumonia syndrome, which has an extremely large differential diagnosis including infectious and noninfectious entities, usually does not require empiric therapy but does mandate a systematic and often invasive approach to establishing the diagnosis.

Acute Community-Acquired Pneumonia

All patients with a clinical diagnosis of pneumonia leading to ICU admission should have a blood culture, a Gram stain, and culture of lower respiratory tract secretions. If a significant pleural effusion is present it should be aspirated. In nonintubated patients, the immediately available lower respiratory tract sample is sputum, which is a mixture of materials from the lung and the upper respiratory tract and that has many well-known limitations in diagnosis: The patient may not be able to produce a good specimen; the specimen may contain potentially pathogenic organisms that come from the upper respiratory tract but are not the cause of the pneumonia; and contaminating organisms from the upper airway may overgrow the real pathogen on the culture plate and prevent its detection. Despite these problems, sputum examination remains a good place to start, mainly because in a significant minority of cases, and particularly in the most severe cases of bacterial pneumonia, the Gram stain, with subsequent confirmation by culture, can be diagnostic.

The utility of the Gram stain can be maximized by microscopic evaluation of the degree to which the sample is contaminated by upper respiratory secretions.26 A specimen with large numbers of polymorphonuclear leukocytes, more than 25 per high-power field, and few large epithelial cells, fewer than 10 per high-power field, is reasonably likely to represent a good-quality lower respiratory tract specimen and can usefully be examined further to determine the predominant bacterial morphotype present. Those specimens with large numbers of epithelial cells and few leukocytes are mostly saliva; they are unlikely to yield useful diagnostic information and should be discarded after another and better specimen has been obtained. In a patient with a clinical presentation consistent with acute bacterial pneumonia, a microscopically acceptable specimen that contains large numbers of bacteria of a single morphotype and few bacteria of other types is strong evidence that this is the cause of the pneumonia. A fairly definitive result such as this occurs in fewer than 33% of cases; however, it occurs more commonly in the most severe cases—those with extensive pneumonia caused by S. pneumoniae, S. aureus, and enteric facultatively aerobic gram-negative bacilli—and in cases in which an endotracheal tube has been placed, which permits the collection of a less heavily contaminated lower respiratory tract specimen.

Preliminary results of blood and sputum cultures are usually available at 24 to 48 hours. Whereas a positive blood culture for a recognized pneumonia pathogen is definitive, sputum cultures must be interpreted in light of the original Gram stain of the specimen, bearing in mind that potential pathogens grown from a specimen that contained mainly upper respiratory tract material may not be the cause of the pneumonia and that, if the patient received antimicrobials before providing the specimen, susceptible pathogens such as S. pneumoniae might not be detected.

In most cases of acute bacterial pneumonia, the investigations described above are all that is needed. Patients with severe bacterial pneumonia leading directly to ICU admission very frequently have positive blood or respiratory secretion cultures, whereas those who are less ill at the time of hospital admission usually improve with empiric antimicrobial therapy (whether or not a positive culture is obtained). Further investigation is needed in patients whose condition deteriorates or fails to improve and in whom the initial investigation has failed to provide an etiologic diagnosis to guide changes in therapy. The first step in a newly intubated patient is to repeat the Gram stain and culture of lower respiratory tract secretions. In deteriorating nonintubated pneumonia patients, the procedure of choice is fiberoptic bronchoscopy using a protected brush to collect a specimen from the consolidated lung segment or bronchoalveolar lavage (BAL) of the region. Because many of these patients have significantly compromised respiratory function, the procedure should be performed by an experienced bronchoscopist with other personnel present to continually monitor the patient's respirations, blood pressure and pulse, and oxygenation. A pulse oximeter to monitor arterial oxygen saturation continuously is mandatory. In patients who have significant arterial hypoxemia to begin with, it is usually best to insert an endotracheal tube and provide supplemental oxygen and assisted ventilation to facilitate bronchoscopy. The specimens obtained should be Gram stained and cultured quantitatively for conventional bacteria; the finding of more than 105 colony-forming units/mL from a BAL specimen (or 103 colony-forming units/mL from a protected brush) of a recognized pathogen is diagnostic of infection.27–29 The specimen also should be stained and cultured for Legionella species, mycobacteria, and fungi and, where available, cultured for viruses. In most cases, it is also wise to do cytologic examination of the specimen.

Among the major pneumonia pathogens not detectable by routine Gram stain and culture of respiratory secretions, the one with the greatest propensity to present with an illness indistinguishable from usual acute bacterial pneumonia is Legionella pneumophila. For this reason, some hospital laboratories in areas of high prevalence routinely culture all submitted lower respiratory tract secretions for this organism. If this culture is not done routinely, it should be ordered specifically for all patients in whom an etiologic diagnosis is not immediately evident from the Gram stain. A direct fluorescent antibody stain for Legionella is also available. This can be done on sputum, endobronchial aspirates, or bronchoscopy specimens, but, particularly with sputum, the sensitivity of the test is not high, and false-positive results can result from cross-reaction of the fluorescent antibody with other gram-negative species. A urinary antigen detection test for Legionella is available in many centers and has excellent specificity; however, particularly early in the course of the illness, the sensitivity is not sufficiently high to exclude the diagnosis.16

In cases of severe pneumonia that resist diagnosis, particularly when the clinical course suggests the possibility of infection with an “atypical pneumonia “ pathogen, additional investigations should be ordered according to the epidemiologic features of the case. In appropriate cases, culture of respiratory secretions and stool for viruses is worthwhile. Acute and 2- to 4-week blood specimens for serologic examination will establish the diagnosis in M. pneumoniae, Chlamydia speceis, C. burnetii, and the major viruses causing pneumonia (influenzae A and B, adenovirus, RSV, and CMV). Single high-titer–specific antibody levels can also be used to make most of these diagnoses earlier in the clinical course. Serology is also valuable in diagnosing some rare bacterial infections, such as Francisella tularensis pneumonia and brucellosis. Legionnaire's disease also can be detected serologically, but, because the antibody response to the organism may be slow, convalescent blood specimens should be drawn at 2 weeks and 6 weeks.

Direct enzyme immunoassay antigen detection performed on respiratory secretions is widely available for diagnoses of influenza A and RSV and is very useful in establishing a diagnosis quickly. At the time of this writing, rapid diagnostic testing for the SARS coronavirus is still under development; the diagnosis still relies on clinical case definitions associated with investigation of the outbreak (probable case: exposure to a known case or travel to an area of local transmission, fever, respiratory symptoms, and pulmonary infiltrate on chest radiograph without an alternative diagnosis).

Suspicion of PCP, generally based on the clinical and radiographic features and history of factors placing the patient at increased risk of HIV infection, should prompt early consideration of fiberoptic bronchoscopy with BAL. Serologic testing for HIV is usually also a part of this investigation.

Aspiration Pneumonia

The approach to aspiration-associated acute bacterial pneumonia caused by the usual pyogenic aerobic organisms is the same as that for pneumonia caused by subclinical aspiration. However, most aspiration CAPs, particularly those progressing to lung abscess, are caused by organisms constituting the normal flora of the oropharynx. These are predominantly anaerobic. Culture of sputum or endotracheal aspirates will always demonstrate some of these organisms because of the inevitable contamination of these specimens with saliva, and the microbiology laboratory will report the result of aerobic cultures as containing “normal flora.” Anaerobic culture of such specimens is not worthwhile because they also would demonstrate “normal flora” in virtually all patients whether or not these organisms were pathogenically important. When the clinical presentation suggests aspiration pneumonia and conventional sputum or endobronchial aspirate cultures show only “normal flora,” in most cases it is reasonable to make a clinical diagnosis of pneumonia caused by mixed oropharyngeal anaerobic bacteria and to treat accordingly. In less obvious cases, in which bronchoscopy is done to establish an etiologic diagnosis, it is reasonable to perform aerobic and anaerobic cultures on protected brush or BAL specimens because they are less likely to be contaminated by saliva. Bronchoscopy at some time in the treatment course is indicated in many patients with anaerobic pneumonia or lung abscess, particularly those with no clearly increased risk of significant aspiration, to exclude a predisposing endobronchial lesion.

Chronic Pneumonia

Some of the many infectious and noninfectious diseases that may present as a chronic pneumonia are listed in Table 51-2. Patients with an undiagnosed chronic pneumonia are less common in the ICU environment because the chronicity of progressive symptoms usually leads patients to seek medical attention before intensive care is required. In those who do present in respiratory failure, there are usually extensive bilateral pulmonary infiltrates, and the radiologic pattern is sometimes helpful in suggesting the diagnosis. However, almost every “characteristic” radiologic pattern can be caused by several different disease processes.

Among infectious causes of chronic pneumonia, reactivating pulmonary fibrocaseous tuberculosis generally demonstrates characteristic upper lobe predominance, with cavities and pleural thickening and scarring, but only rarely causes significant hypoxemia, generally when spillage of infected material from tuberculous cavities into otherwise normal lung produces acute airspace inflammation termed acute tuberculous pneumonia (Fig. 51-8). Somewhat less rarely, miliary tuberculosis or chronic hematogenous tuberculosis usually presents with a diffuse miliary or reticulonodular pattern on a chest radiograph (Fig. 51-9), often with respiratory failure, particularly when the disease produces a systemic inflammatory response complicated by the ARDS.30 Infections that can mimic upper lobe tuberculosis reactivation include chronic necrotizing bacterial pneumonia (usually caused by K. pneumoniae, occasionally by P. aeruginosa or S. aureus), melioidosis, and some fungal infections. Miliary or reticulonodular disease (with or without ARDS) causing respiratory failure also occasionally occurs with fungal pneumonias, in particular blastomycosis31 (Fig. 51-10), malignancy, and noninfectious granulomatous diseases.

The initial step in the investigation of a patient with a chronic pneumonia is recognizing that the pace and pattern of the disease do not fit with one of the acute bacterial or atypical pneumonias. Appropriately directed, additional history taking for exposure to tuberculosis or residence in an area endemic for one of the dimorphic fungi can be obtained, and extrapulmonary symptoms or signs associated with the noninfectious diseases listed in Table 51-2 can be sought. Lower respiratory tract secretions should be sent for routine Gram stain and culture for bacteria, and several specimens should be examined and cultured for mycobacteria and fungi. In patients who have spent time in the Far East, examination for ova and parasites should be done to exclude infection with Paragonimus westermani. Cytologic studies for malignancy should be done in most cases. If sputum or endobronchial secretions are used, these should be sent repeatedly (three to five specimens) to maximize yield. In patients from whom satisfactory secretions cannot be obtained, fiberoptic bronchoscopy with BAL is most satisfactory.

Other noninvasive tests for infectious agents, such as serologic testing and skin tests, which are often of some diagnostic value in less severely ill patients, are not helpful in critically ill patients. However, for some of the noninfectious diseases, in particular systemic vasculitides and other immunologically mediated diseases with pulmonary involvement, testing for the specific autoantibodies associated with these syndromes can be very useful (see Chap. 104).

Many of the causes of the chronic pneumonia syndrome also cause extrapulmonary manifestations. The nature of these associated findings is frequently of considerable help in suggesting the correct diagnosis; examples include skin or bone involvement with pulmonary blastomycosis, sinusitis with Wegener granulomatosis, lymphocytic meningitis with coccidioidomycosis or tuberculosis, granulomatous oral lesions with paracoccidioidomycosis, and glomerulonephritis with Goodpasture syndrome and other immunologically mediated systemic diseases. When other organ systems are involved, it is often less risky for a patient with significantly compromised respiratory function to undergo biopsy of the extrapulmonary site of involvement rather than of the lung.

If a diagnosis is not promptly established by the initial examination of lower respiratory tract secretions, if the clinical constellation of findings is not characteristic enough to establish a diagnosis on these grounds, and if no extrapulmonary site of involvement suitable for biopsy is present, most critically ill patients with chronic pneumonia immediately should undergo open lung biopsy. The range of potential diagnoses and risks associated with empiric therapy for most of the diagnostic entities is simply too large to attempt management without establishing a definitive diagnosis. In a patient who has reached a stage of the disease requiring intensive care, the necessary time for further observation of the course or attempts at empiric trials of therapy is lacking.

Antimicrobial Therapy

Acute Community-Acquired Pneumonia

When a specific etiologic diagnosis for pneumonia has been established on the basis of the blood culture or Gram stain and culture of secretions from the respiratory tract, antimicrobial therapy directed specifically at that pathogen can be prescribed (Table 51-3). The regimens suggested in this chapter are applicable when in vitro testing demonstrates that the organism is susceptible to the listed antimicrobial agents; resistant strains would require appropriate alternative therapy. However, in most cases of pneumonia, the etiologic diagnosis is not available at the outset, and empiric therapy must be directed at the most probable microbial etiologies in the particular patient being treated. Because patients requiring intensive care are usually in a relatively precarious state that makes the consequences of undertreatment unacceptable, it is a principle of therapy in this class of patient that antimicrobial treatment should be very conservative; that is, it should cover all etiologic possibilities except the quite remote ones. Some suggested empiric regimens are presented in Table 51-4; these are loosely congruent with, but not identical to, some recent recommendations from expert panels1–3 that thus far have been unable to reach a consensus, mainly because definitive comparative studies of antimicrobial regimens do not exist.

For most cases of severe acute pneumonia, a third-generation cephalosporin combined with quinolone or a macrolide is appropriate because it works against most conventional bacteria likely to be acquired in the community and atypical pathogens such as Legionella and Mycoplasma. A respiratory quinolone or macrolide alone is not advised in the more severely ill ICU population because of the emergence of significant resistance to these agents in Streptococcus pneumoniae and some other species.

When the clinical context suggests the possibility of infection caused by C. burnetii or C. psittaci, a tetracycline such as doxycycline 100 mg intravenously (IV) every 12 hours should be included in the regimen. Rifampin 600 mg IV or orally can be added for patients with fulminant atypical pneumonia and for patients with pneumonia caused by L. pneumophila or Q fever that is responding poorly to the initial treatment regimen.

Aspiration Pneumonia

Most cases of witnessed or otherwise recent aspiration of oropharyngeal or gastric contents associated with a new pulmonary infiltrate represent an acute chemical pneumonitis rather than an infection. These cases usually resolve without antimicrobial therapy, with deep breathing and coughing, or with chest physiotherapy, when required. There is no convincing evidence that early administration of antimicrobials decreases subsequent incidences of complicating bacterial pneumonia, although there is evidence that such treatment may be associated with subsequent pneumonia caused by a relatively antimicrobial-resistant organism.24,25 Aspiration that has resulted in pneumonia, lung abscess, or empyema caused by oropharyngeal anaerobic bacteria has usually been treated, at least initially, with penicillin. However, in a critically ill patient with this syndrome, therapy should usually begin with penicillin 2 million U IV every 4 hours and metronidazole 750 mg IV every 6 hours or with clindamycin 900 mg every 8 hours. Aspiration can also be complicated by later development of aerobic acute bacterial pneumonia; in these cases, the regimens listed in Tables 51-3 and 51-4 apply. Much less commonly, aspiration can produce an acute, rapidly progressive necrotizing pneumonia. This infection usually involves anaerobes and facultatively aerobic enteric gram-negative bacilli. The clinical context is often major aspiration in a patient without gastric acid (and therefore a large gastric bacterial population) or with an accumulation of feculent material in the stomach caused by adynamic ileus, upper GI bleeding, or bowel obstruction. Treatment must include adequate coverage for anaerobes, including Bacteroides fragilis, and aerobic gram-negative organisms. As in other acute gram-negative pneumonias, it is unwise to rely on an aminoglycoside alone for aerobic gram-negative coverage. Acceptable initial regimens include piperacillin/tazobactam 3.375 g IV every 6 hours with gentamicin 4.5 mg/kg every 24 hours, clindamycin 900 mg IV every 8 hours with cefotaxime 2 g IV every 6 hours, and meropenem 1 g every 8 hours with gentamicin 4.5 mg/kg IV every 24 hours. Ciprofloxacin can be substituted for the aminoglycoside when toxicity is a concern.

Chronic Pneumonia

Empiric therapy is seldom indicated for patients presenting with a chronic pneumonia. The pace of the illness is generally such that a delay of hours or a few days in establishing the diagnosis is not critical, and the large differential diagnosis makes selection of empiric treatment difficult. One occasional exception to this rule is extensive pneumonitis or pulmonary hemorrhage in a patient with sufficient extrapulmonary evidence to support a clinical diagnosis of a systemic inflammatory disease such as lupus erythematosus, Goodpasture syndrome, polyarteritis, or Wegener granulomatosis (see Chap. 104).

A full discussion of specific therapies for all the infectious and noninfectious causes of chronic pneumonia is beyond the scope of this chapter; however, treatment regimens for the more common fungal pneumonias are listed in Table 51-5. Drugs used in the treatment of critical illness or respiratory failure caused by tuberculosis are listed in Table 51-6. In general, in seriously ill patients, use of four antituberculous drugs initially (isoniazid, rifampin, pyrazinamide, and streptomycin; ethambutol substituted for streptomycin, if necessary, or added if drug resistance is suspected) is recommended.32 Addition of a corticosteroid (methylprednisolone 125 mg IV every 12 hours) can also be considered in the nonimmunocompromised patient with severe tuberculosis in an attempt to speed resolution of the exudative inflammatory response, but controlled studies to support this practice are lacking.

Supportive Therapy

In most respects, supportive therapy for patients with pneumonia resembles that for other patients with infections requiring intensive care. The most frequent reason for a patient with pneumonia to require intensive care is respiratory failure. An approach to management of this problem is detailed in Chapter 38. Most patients have intravascular volume depletion at presentation and require some intravenous volume expansion with crystalloid solutions. However, septic pneumonia patients are also at increased risk of developing more generalized pulmonary edema, as in ARDS (see Chap. 38). Hypotension that does not respond to reasonable volume expansion is usually due to septic shock, the management of which is detailed in Chapter 46.

As in other critically ill patients, a multiplicity of metabolic, hematologic, and blood electrolyte abnormalities occur and may require correction. Especially important in the pneumonia patient is hyponatremia caused by inappropriate vasopressin secretion, leading to impaired water excretion by the kidney; this is common and makes administration of hypotonic intravenous solutions potentially hazardous.33 The correction of metabolic alkalosis of any cause also may be important because this may be associated with an improvement in arterial oxygenation caused by improved ventilation/perfusion matching related to potentiation of regional hypoxic pulmonary vasoconstriction.34

Because patients with acute pulmonary infections frequently shed potentially pathogenic organisms into their immediate environment, the risk of exposure of caregivers and other patients to these pathogens must be considered. Many of the common acute respiratory pathogens causing pneumonia are normal flora of the respiratory tract (e.g., S. pneumoniae, H. influenzae, and Moraxella catarrhalis) and therefore require no special precautions during patient care. However, some bacterial pathogens do pose significant risk to other patients if transmitted via aerosol or on the hands or equipment of caregivers. These include S. aureus (especially methicillin-resistant S. aureus) and multiply-resistant gram-negative bacilli. When these organisms are cultured or suspected patients should be cared for in separate rooms, where possible, using appropriate infection control precautions35 (see Chap. 4). Some viral infections also pose a risk to staff and other patients. Examples include influenzae A and B (for which staff vaccination is highly effective), RSV, and acute Varicella zoster.

During the SARS outbreak of March 2002, transmission of infection to caregivers accounted for the majority of cases in some centers, with several fatalities among those infected. Most transmissions occurred by direct contact or by respiratory droplet spread from undiagnosed infected patients to staff not using personal protective precautions. When active case identification was put in place, potential cases were rapidly put in isolation, and caregivers were protected by using N95 masks, eye protection, gloves, gowns, and strict hand washing for all patient contact, transmission was quickly halted. Patients requiring ventilatory support in the ICU were particularly important vectors of infection, mainly related to droplet spread of the virus during endotracheal intubation and other respiratory care procedures. The following measures to minimize risk to staff during intubation for SARS-related respiratory failure are suggested:

1. Anticipate the need for ventilatory support to ensure that intubation occurs in a well-prepared, nonemergent controlled setting.

2. Intubation should be performed by the most skilled individual available.

3. Minimize staff in procedure room.

4. Intubate in a closed-door, well-ventilated, and preferably negative-pressurized room.

5. Minimize coughing and other droplet-generating respiratory efforts by the patient by using good preintubation sedation; consider using muscle relaxation to facilitate intubation.

6. Personnel in the room should have N95 masks checked for correct fit, face shields, gowns, and gloves.

7. Meticulous hand washing should occur after any procedure, particularly after removing protective clothing.

8. Surfaces in the procedure room should be disinfected before the area is used for another purpose.

9. Continue these infection control procedures during subsequent ICU care for respiratory failure.

Nosocomial pneumonia differs significantly from CAP in aspects of pathogenesis and in the range of usual microbial pathogens seen.36,37 Approaches to diagnosis are generally similar to those for CAP; however, the approach to antimicrobial therapy differs considerably, reflecting the different spectrum of pathogens. In addition, because nosocomial pneumonia by definition occurs in patients already under medical care for illnesses requiring hospitalization, the possibility of preventing its occurrence exists and is an important aspect of the day-to-day care of the critically ill patient (see Chaps. 4 and 10).

Pathogenesis and Microbiology

The most common microbial causes of nosocomial pneumonia are listed in Table 51-7. This table was constructed from data gathered in a number of surveys of different patient populations, and it must be emphasized that hospital populations are not identical with respect to risk for pneumonia caused by each of these pathogens. As in CAP, the basic event leading to pneumonia is usually subclinical aspiration of oropharyngeal material, an event for which hospitalized patients are at increased risk on account of disease and medical intervention. These increased risks include depressed level of consciousness caused by head injury, metabolic disease, or sedative and anesthetic drugs, and impaired laryngeal reflexes and esophageal sphincter function caused by neurologic disease or instrumentation of the airway or esophagus. Although endotracheal intubation with a cuffed tube may prevent frank aspiration of gastric contents, it is no barrier to the minor degrees of aspiration required to transfer potential pathogens from the upper to the lower respiratory tract. The patient's ability to deal effectively with contamination of the lower respiratory tract may also be compromised. Mucociliary clearance may be impaired by damage to the tracheal mucosa by endotracheal tubes, metabolic disease, or drugs; immunologic defenses may be impaired by underlying disease or by immunosuppressive therapy.

The explanation for the nature of the common pneumonia pathogens is the pattern of colonization of the upper airway, which determines which organisms gain access to the lung when aspiration occurs. In previously well patients admitted to ICUs with acute illness that compromises airway defenses, pneumonia in the few days after admission is caused most often by organisms that are frequent colonizers of the upper airway in normal persons: S. aureus,Streptococcus species, and, to a lesser extent, H. influenzae and M. catarrhalis.38 Patients with underlying chronic lung disease are frequently colonized with these same pathogens, but nontypeable H. influenzae and M. catarrhalis are more frequent in patients with underlying lung disease than in normal persons and therefore cause early nosocomial pneumonia more frequently in these patients.39,40 Patients who have other chronic debilitating diseases or those who have been in the hospital for a number of days are at increased risk of having upper airway colonization and subsequent pneumonia caused by enteric facultatively aerobic gram-negative bacilli such as K. pneumoniae, E. coli, and Enterobacter species, whereas patients with underlying structural lung disease, the immunosuppressed, and those who have been treated with antimicrobials or who have had a lengthy stay in an ICU are at increased risk for colonization or infection with more antimicrobial-resistant hospital-acquired pathogens such as P. aeruginosa, Stenotrophomonas maltophilia, or Acinetobacter species.41 The source of staphylococci or Enterobacteriaceae causing these infections is most often the patient's own endogenous microflora; however, many of the more resistant pathogens may be transmitted to patients by contact spread from the hands of caregivers or, particularly with water-borne organisms such as Pseudomonas or Acinetobacter, by contaminated respiratory therapy equipment.

Among the pathogens in the atypical pneumonia group, only viruses and L. pneumophila are notable causes of nosocomial pneumonia. Influenza A and RSV are sporadically reported to cause outbreaks of pneumonia in the hospital, generally by respiratory or hand-to-hand spread from other infected individuals, and there is reason to believe that CMV can also occasionally cause pneumonia leading to respiratory failure in immunocompetent hospitalized patients; the pathogenesis of nosocomial infection in this population is unknown. In the March 2002 SARS coronavirus outbreak, much of the disease transmission was nosocomial, mainly by direct contact or droplet spread from patients with undiagnosed SARS to health care workers not using any personal protection against infection, and from them to other patients by the same mechanisms.21 For nosocomial infection with Legionella species, the pathogenesis is inhalation of aerosols containing the organism. In outbreak situations, this can usually be traced to presence of the organism in hospital water supplies, with dissemination by showers and baths and the like, or to cooling tower contamination. Ground excavation near an outside air intake has also been implicated. It is likely that many hospitalized patients are exposed to Legionella, given its widespread presence in hospital water supplies; however, most cases continue to occur among the immunosuppressed hospitalized population, particularly those on corticosteroid treatment for prevention of organ transplant rejection. This is presumably because a patient with suppressed cell-mediated immunity has a reduced capacity to clear even small inhaled inocula of this intracellular pathogen, with which a patient with intact cell-mediated immunity can cope easily.

Diagnosis

The clinical features and radiographic findings associated with nosocomial pneumonia are similar to those of CAP but may be greatly confounded by other disease processes common in hospitalized patients, particularly those in intensive care. The diagnosis is suggested by the presence of a pulmonary infiltrate consistent with pneumonia on chest radiograph, increased amount and purulence of lower respiratory tract secretions (sputum or endobronchial aspirates), deterioration in gas exchange and increased dyspnea, and an increase in temperature or leukocyte count. There are problems with the sensitivity and specificity of this clinical definition. Postmortem studies of critically ill patients with extensive lung disease of other causes have indicated that many patients have clinically unsuspected bacterial pneumonia at the time of death. However, there is evidence that, in the general ICU patient population, other processes, such as chemical pneumonitis caused by aspiration, atelectasis, and purulent tracheobronchitis, can produce many of or all the same clinical findings as pneumonia, and that pneumonia can be significantly overdiagnosed in this context.42 In intubated, ventilated patients in particular, quantitative bacteriologic evaluation of bronchoscopic specimens of secretions from the lower respiratory tract has been advocated as a way of overcoming this problem, and this approach is discussed comprehensively in Chapter 43.

For most patients, a careful clinical assessment combined with conventional microbiologic studies for the diagnosis of nosocomial pneumonia in ward patients and in intubated ICU patients remains a reasonable first approach. Some patients have clinical presentations with such marked pyrexia, leukocytosis, deterioration in lung function, and advancing infiltrates that the diagnosis is fairly certain. It remains only to obtain the appropriate specimen to establish the etiology; in most such cases, this is sputum or endobronchial aspirates, a blood culture, and needle aspirate of pleural fluid if a significant amount is present. It is mainly in patients with less florid presentations, with marginal changes in the major diagnostic parameters and with infiltrates that could be atelectasis or pneumonia, in whom the diagnosis is substantially in doubt. In these cases, the usual best policy is to obtain endobronchial secretions for Gram stain and culture, provide chest physiotherapy and or frequent posturing of the patient and endobronchial suctioning, and withhold antimicrobial treatment initially. Over the ensuing hours to days, the result of the culture will become available; the additional information regarding the progress of the pulmonary process, the systemic inflammatory response without antimicrobial treatment, the presence and quantitation of leukocytes and bacteria on Gram stain, and the results of initial cultures will aid in making the judgment about whether an infection requiring antimicrobial treatment is present. In cases in which substantial uncertainty remains, fiberoptic bronchoscopy with a protected specimen brush or BAL with quantitative bacteriology is a reasonable next step.

Management

The principles of supportive management for nosocomial pneumonia are identical to those outlined for CAP. However, there are several differences in antimicrobial therapy because of the considerably different range of etiologic agents seen. Atypical pneumonia is much less common in hospitalized patients and, when it occurs, is most often caused by Legionella species. Bacterial pneumonias make up the vast majority of cases and are much more likely to be caused by S. aureus or enteric gram-negative bacilli and, in particular subgroups of patients, by relatively antimicrobial-resistant gram-negative bacilli. Suggested empiric antimicrobial regimens for hospital-acquired pneumonia are listed in Table 51-8 and those for specific proven pathogens are listed in Table 51-3.

The pathogens causing hospital-acquired pneumonia are much more likely to cause a destructive necrotizing pneumonia than are the common community-acquired pathogens and therefore are more likely to result in formation of abscesses and empyema, pneumothorax, and permanent fibrosis of involved lung regions. These pathogens also are more difficult to eradicate permanently from the lung with antimicrobial therapy. Accordingly, duration of antimicrobial therapy is generally longer, usually at least 3 weeks to as many as 6 or 8 weeks, depending on the initial response, presence of abscess cavities, and susceptibility of the organism. Treatment should always be parenteral initially and for most enteric gram-negative organisms should usually include two different agents effective against the pathogen, generally a β-lactam drug and an aminoglycoside or a quinolone. After 7 to 10 days, if a good clinical response has been observed, completion of the treatment course with an appropriate oral agent is reasonable. Oral agents considered suitable for gram-negative pneumonias include cotrimoxazole, amoxicillin/clavulanic acid, cefuroxime axetil, cefixime, and, for resistant organisms such as Pseudomonas species, ciprofloxacin.

As in outpatients, pulmonary aspiration in hospital patients can result in a simple chemical pneumonitis or atelectasis that does not require antimicrobial therapy. Hospitalized patients are much more likely to have been receiving antimicrobial agents for other indications, agents that decrease gastric acidity, and are more likely to have conditions predisposing to development of large numbers of pathogenic bacteria in the oropharynx and the stomach (e.g., recent surgery, adynamic ileus, or tube feeding). This increases the risk of an acute necrotizing bacterial pneumonia in these patients. Accordingly, if there is evidence to support an active infectious process after presumed or witnessed aspiration, antimicrobial therapy directed at GI anaerobes and relatively resistant enteric gram-negative bacilli should be given.

Antimicrobial selection for hospital-acquired pneumonia, to a much greater extent than that for CAP, must be based on knowledge of which pathogens are prevalent in the environment in a particular hospital or even in a particular unit within the hospital, because antimicrobial resistance patterns vary greatly in different locations. The recommendations listed in the tables must therefore be used with caution and with knowledge of local conditions.

Prevention

The considerable morbidity rate associated with nosocomial pneumonia, particularly in critically ill patients, has led to much interest in developing methods to prevent these infections.43,44 Prevention would clearly depend on breaking the pathogenetic chain leading from colonization of the upper respiratory and GI tracts with potential pathogens to aspiration of these colonizers into the lower respiratory tract; establishment of lower respiratory tract colonization; and then failure of the pulmonary clearance mechanisms and immune system to contain the organism, leading to invasion of the pulmonary parenchyma. Attempts to break the chain at each of the points mentioned have been tried by various investigators, with limited success.

Minimizing the potential for the transmission of relatively antimicrobial-resistant pathogens between patients clearly is the first priority. Basic measures to accomplish this include hand washing before and after patient contact by all caregivers, avoiding the sharing of bedside or respiratory therapy equipment between patients, and isolation of patients known to be infected or colonized with multiply-resistant bacterial pathogens.

Interventions to prevent upper airway and gastric colonization by enteric gram-negative bacilli have recently been tried in many centers. Based on evidence that stomach acid helps minimize such colonization and that administration of H2 blockers or antacids for prophylaxis against upper GI bleeding increases colonization, use of sucralfate, an agent that protects the mucosa and prevents bleeding without increasing stomach pH, has been advocated, and studies from some centers have been demonstrated that its use decreases the rate of nosocomial pneumonia in intubated patients. However, others have been unable to confirm that gastric colonization is a risk for airway colonization and pneumonia or that use of sucralfate rather than antacids or histamine blockers is protective. Although the controversy on this issue continues, it should be pointed out that sucralfate is cheap, nontoxic, and probably as effective as other agents, so if bleeding prophylaxis is used, sucralfate may be preferable to an H2 blocker or antacid, except in patients at special risk for upper GI bleeding. Other suggested strategies to limit bacterial proliferation in the stomach and subsequent spread to the respiratory tract include using small bowel feeding tubes instead of gastric tubes, intermittent feeding to maintain gastric acidity, and acidification of feeds, but none of these has been convincingly shown to be effective.

Different prophylactic antimicrobial regimens, topical and systemic, have also been used to prevent colonization and pneumonia. These regimens have often been shown to be effective in limiting the incidence of pneumonia in ICU patients, but there continues to be concern that their use may eventually lead to an increase in the prevalence of antimicrobial resistance. Also, in most reported studies, the decreased rate of pneumonia was not associated with a decrease in mortality rate or in use of ICU resources. For these reasons, none of the currently available prophylactic antimicrobial regimens can be recommended.

Decreasing the risk of aspiration in seriously ill patients involves mainly commonsense measures. For a nonintubated patient who has difficulty in swallowing or decreased level of consciousness, nursing the patient on his or her side with the head slightly down may be helpful. Frequent and careful suctioning and mouth care are also important. In patients with endotracheal or nasogastric tubes and those with otherwise disturbed esophageal sphincter function, the upright position and sleeping with the head of the bed up reduces the risk of gastroesophageal reflux and aspiration. When the nasogastric tube is in place solely for feeding purposes, and it has been established that gastric emptying is adequate, the large-bore tube can be changed to a soft small-bore tube, which is less likely to interfere with the esophageal sphincter and is also more comfortable. In orally intubated patients, use of an orogastric tube rather than a nasogastric tube is also preferred.

For intubated patients, exemplary nursing care can have a substantial role in decreasing the risk of pneumonia. Frequent turning of patients while they are in bed and minimizing the use of deep sedation or muscle relaxants help avoid dependent secretion accumulation in the lungs. Use of oscillating beds to accomplish this has had mixed success with respect to efficacy and tolerance by patients, so their routine use is not currently recommended. Frequent gentle suctioning of the endotracheal tube and careful monitoring of endotracheal cuff seal and pressure are also likely to help minimize secretion accumulation and prevent pneumonia.45 Good mouth care, suctioning of the oropharynx before endotracheal tube cuff deflation or before turning of the patient, no-touch sterile suctioning technique, and early detection and correction of gastroesophageal reflux problems are other measures that have yet to be studied in a systematic way with regard to infection rate; but they almost certainly account for much of the extremely wide variation in nosocomial pneumonia rates in different ICUs.

An active infection control program is also critical to minimizing nosocomial pneumonia risk.46 Careful monitoring of infection rates and the occurrence of pathogens with a propensity for nosocomial spread is needed to detect potential reservoirs for cross-infection of patients and to guide institution of infection control measures. In general, an ICU with a pneumonia rate higher than approximately 10% in patients ventilated for longer than 48 hours, with a sustained increase in pneumonia rate over baseline, or with a predominance of water-borne gram-negative organisms such as Pseudomonas species or Acinetobacter species or any other single species as a cause of most cases of pneumonia, probably has a potentially correctable infection control problem.

Immunocompromised patients with fever and new pulmonary infiltrates usually come to the attention of the intensive care physician when they develop imminent or established acute respiratory failure or when a period of endotracheal intubation with mechanical ventilatory support is needed to facilitate an invasive diagnostic procedure. That both of these events occur relatively frequently says a great deal about the many special problems posed by this group of patients: The differential diagnosis of pulmonary inflammatory processes is wide, establishing a definitive diagnosis is often difficult, empiric antimicrobial therapy is often ineffective, and the patient's condition can deteriorate rapidly.

The nature of the immunologic abnormality that is present in a given case has a major bearing on the nature of the infective processes likely to be present. The major categories of immunologic abnormality include (a) deficient humoral immunity (these are patients with agammaglobulinemia and multiple myeloma or chronic lymphocytic leukemia, without intensive chemotherapy), (b) deficient phagocytic cell function or number (patients with chemotherapy-induced granulocytopenia make up the majority in this group), and (c) deficient cell-mediated immune function (includes patients on corticosteroids or other immunosuppressive drugs for the control of a variety of inflammatory diseases, control of rejection after organ transplantation, and less intensive chemotherapy for a variety of malignancies; it also includes patients with advanced HIV infection, which is discussed in more detail in Chap. 48). Many patients have mixed immune dysfunctions, especially those on rigorous chemotherapeutic regimens for cancer. Table 51-9 lists the major pulmonary infections most strongly associated with each type of immune dysfunction.

Clinical Features

In addition to the information regarding the nature of the immunodeficiency noted above, clues to the etiologic diagnosis may be available from details of the history and physical examination.

The timing of the onset of the illness, with respect to previous chemotherapy and duration of immunosuppression, is often very helpful. After cytotoxic chemotherapy resulting in granulocytopenia, acute bacterial infections predominate in the first 2 to 3 weeks. Subsequently, especially with illnesses occurring in the face of broad-spectrum antimicrobial therapy, opportunistic fungal infections are increasingly common. After bone marrow and other transplantation procedures, bacterial infections also predominate early, with viral and fungal infections occurring weeks to months later. In this patient population, CMV pneumonia is of particular importance, occurring with increased frequency in patients known to be seropositive for CMV; in bone marrow recipients, CMV pneumonitis is also more common in older patients, those who have received total-body irradiation, and those with severe graft-versus-host disease.47,48 Some noninfectious causes of pulmonary infiltrates also follow a somewhat predictable time course: radiation pneumonitis usually occurs 4 to 10 weeks after treatment; pulmonary edema caused by acute toxic myocarditis after high-dose doxorubicin therapy usually occurs a few days to a few weeks after treatment, whereas cardiomyopathy from doxorubicin and daunorubicin is increasingly frequent with increasing cumulative dose; and interstitial pneumonitis caused by bleomycin is common at cumulative doses larger than 450 U, although damaging pulmonary reactions to this cytotoxic drug and to others can occur with much smaller doses.49

Although the pace at which the pulmonary process progresses is to some degree characteristic for each etiologic agent, this is highly variable and greatly dependent on the nature and severity of the underlying immunologic deficit. Aspergillus species, for example, produce an indolent localized pneumonia in patients treated with immunosuppressive drugs and corticosteroids if phagocytic cell function is relatively preserved, but in patients with severe combined immunodeficiency and in some with long-term granulocytopenia from chemotherapy for myelogenous leukemia, the illness may be fulminant.50Pneumocystis carinii causes a severe, rapidly progressive diffuse pneumonia in patients with lymphoma or leukemia but usually is an insidious illness in those with HIV. It is also worth bearing in mind that an apparent change in the pace of progression of disease may signal the presence of more than one process, for example, an acute bacterial infection superimposed on a more slowly progressive disease, such as interstitial pneumonitis.

The physical examination is only occasionally revealing in this patient group. Examination of the chest almost never adds to the radiologic assessment. However, extrapulmonary findings may be very helpful in the differential diagnosis and may provide a potential alternative site for diagnostic testing. Central nervous system involvement (meningitis or encephalitis) indicates systemic fungal infections, tuberculosis, bacteremia with metastatic foci, or disseminated viral infections and mandates computed tomography of the brain followed by lumbar puncture. Disseminated aspergillus infection, in addition to involving the lung and brain, may produce necrotizing skin lesions. Ecthyma gangrenosum signals bacteremia, usually with P. aeruginosa and usually producing a diffuse multinodular pneumonia. Staphylococcal bacteremia and pneumonia may be associated with widespread cutaneous pustules, whereas the rare case of disseminated candidiasis with pulmonary involvement may be associated with a papular rash, myalgias, or “cotton wool” exudates on funduscopic examination.

Radiographic Features

The chest radiographic appearance of the pulmonary infiltrate can be a useful guide to a differential diagnosis but is seldom specific enough to establish the diagnosis. The most important differentiation to be made is whether the infiltrate is a localized or a diffuse process.

Diffuse infiltrates, if they are of infectious origin, imply infection with a pathogen that has reached the lung hematogenously in large numbers or has spread rapidly along the respiratory mucosa early in the course of the infection. Examples include viral infections such as CMV, which is actually a systemic infection; hematogenous bacterial pneumonia (P. aeruginosa most commonly); miliary or chronic hematogenous tuberculosis; and candidemia with secondary candidal pneumonitis. PCP also is typically a diffuse pneumonia (Fig. 51-11). Noninfectious processes producing diffuse infiltrates are those in which the entire lung is exposed equally to the damaging agent through the bloodstream or the lymphatics (Table 51-10). Examples include drug-associated lung injury, leukemic infiltration or lymphangitic carcinomatosis of the lung, and idiopathic interstitial pneumonitis.51

Localized infiltrates caused by infections are generally those in which the organism has gained access to the lung by aspiration of oropharyngeal material or by inhalation of the pathogen. Examples include most cases of acute bacterial pneumonia, including nocardial pneumonia and L. pneumophila pneumonia, and most cases of opportunistic fungal pneumonia, such as those caused by Aspergillus, Mucor, and Cryptococcus species. The appearance of the infiltrate sometimes provides a diagnostic clue. Legionella pneumophila often produces airspace consolidation with a peripheral, ill-defined, rounded density, often with rapid subsequent progression (see Fig. 51-4).16 Localized infiltrates caused by Aspergillus usually begin as enlarging nodular infiltrates or as patchy bronchopneumonia in the granulocytopenic patient (Fig. 51-12), with the development of central necrosis, cavitation, and “air crescent” signs associated with marrow recovery.52 Hematogenous spread of infection causing a localized infiltrate also occurs occasionally, although this most often produces multiple, locally spreading infiltrates that may not be anatomically adjacent; hematogenous bacterial infections, especially septic thromboembolia, are examples of this. The major noninfectious entities producing localized infiltrates are atelectasis, pulmonary embolism with infarction, and metastatic neoplastic disease.

Approach to Diagnosis and Management

Diagnosis and management must be considered together, particularly for patients requiring support in an ICU. There are usually two phases to the approach. In the first phase, a presumptive diagnosis or group of potential diagnoses is formulated and empiric therapy is begun while relatively noninvasive diagnostic testing is implemented. Many patients respond to initial therapy or, if they do not, have a tenable diagnosis established on the basis of the initial investigation, leading to a revision of the initial therapeutic regimen. In the second phase, which is marked by the increasingly clear need for a definitive diagnosis because of the failure of empiric therapy or the pace and severity of progression of disease, increasingly invasive diagnostic testing is performed until the diagnosis is established and definitive therapy begun. The key to appropriate management is to take great care to neither abandon the frequently correct initial diagnosis and treatment too soon (thereby subjecting many patients to the risk of an unnecessary lung biopsy) nor wait too long to proceed to a definitive diagnostic procedure (making the diagnosis too late to reverse the disease).

The major considerations governing the initial empiric antimicrobial approach and the selection of diagnostic procedures are (a) the underlying disease and the timing of the pulmonary process with respect to it, (b) the presence of major clues to the diagnosis from the initial clinical evaluation, (c) the radiologic pattern, and (d) the physiologic stability of the patient and the rate of progression of the disease. Patients with pulmonary infiltrates that fit a commonly seen clinical and radiologic pattern quite well and who are physiologically stable can be treated empirically pending the results of initial diagnostic testing. Those who have less certain clinical and radiologic patterns and who are unstable or getting worse quickly require prompt and definitive diagnostic testing.

Patients with diffuse infiltrates should receive intravenous wide-spectrum antimicrobials directed at hematogenous bacterial pneumonia, at least until initial culture results are available. Piperacillin/tazobactam 3.375 g IV every 6 hours and ciprofloxacin 400 mg IV every 12 hours is usually acceptable. Cotrimoxazole (20 mg/kg trimethoprim with 100 mg/kg sulfamethoxazole per day in four divided doses) may be added to cover P. carinii (unless the patient has been receiving cotrimoxazole prophylaxis, in which case the diagnosis is essentially excluded). Investigations include blood cultures, respiratory and stool specimens for viral culture, culture of blood for CMV, respiratory secretions for Gram stain and culture, and examination for acid-fast bacilli and fungi. In patients with predominantly cell-mediated immune impairment (lymphoma, steroid, or immunosuppressive drug therapy or organ transplant) who have not received cotrimoxazole and have negative blood cultures, PCP is the leading infectious possibility. Unlike what is seen in HIV patients with PCP, the number of organisms in the lung is not large; hence sputum examination for the organism is not useful. Early bronchoscopy with BAL or a protected specimen brush is advised.53 The specimen obtained should be examined for the full range of pathogens mentioned above. Cytologic examination of cells for evidence of CMV, culture of respiratory secretions, and blood for CMV will help in establishing that diagnosis. Bronchoscopic methods have excellent diagnostic yield for infectious diagnoses in this setting but are somewhat less useful in defining noninfectious etiologies for diffuse infiltrates.54 Accordingly, for patients who are at lesser risk for PCP and CMV, for those in whom the initial investigation excludes these possibilities, and for those with rapidly progressive diffuse lung disease, open lung biopsy should be considered. Open lung biopsy remains controversial in this setting: Although a definitive diagnosis can be established in most patients and therapy can be adjusted appropriately, in severely ill patients this often does not lead to survival. Further, significant morbidity caused by the procedure is not uncommon. Nevertheless, there is evidence that in selected cases open biopsy can be very useful by offering the best chance of survival to the subset of patients with reversible disease.55,56 In the face of progressing disease, if a biopsy is to be done, it is important to make the decision to proceed before the patient has progressed to frank respiratory failure.

The diagnosis of localized pulmonary infiltrates caused by conventional bacterial pneumonia can usually be established with sufficient certainty by blood cultures, Gram stain and culture of respiratory secretion and culture, direct fluorescent antibody staining and culture of respiratory secretions for L. pneumophila, and the clinical response to empiric antimicrobial therapy, usually piperacillin/tazobactam and ciprofloxacin, unless it is known that the patient is colonized with a pathogen resistant to these antimicrobials. If the diagnosis is not immediately forthcoming or the patient does not respond promptly to empiric therapy, bronchoscopy should be done and BAL or protected brush specimens sent for examination for bacteria; acid-fast bacilli; direct fluorescent antibody, and culture for Legionella; fungal elements; and cytologic examination. If there is no contraindication, transbronchial biopsy may increase the yield for invasive fungal infection and for malignancy. However, as many as 50% of cases of invasive fungal infections may not be detected by bronchoscopic methods and require open lung biopsy for diagnosis. Culture of aspergilli from nasal scrapings or sputum can be helpful in this regard. Although culture of aspergilli is not indicative of disease in normal persons, in severely immunocompromised patients with granulocytopenia and acute leukemia, such positive cultures are strongly associated with invasive disease.

When the diagnosis has been established, definitive therapy is begun and other agents started empirically are discontinued unless other indications for their use remain. A discussion of the full range of therapies for noninfectious etiologies of pulmonary infiltrates is beyond the scope of this chapter; drug regimens for the major infectious causes of opportunistic pneumonia are presented in Table 51-11.