Harrison's Internal Medicine > Chapter 251. Pneumonia >
Pneumonia is an infection of the pulmonary parenchyma. Despite being the cause of significant morbidity and mortality, pneumonia is often misdiagnosed, mistreated, and underestimated. In the past, pneumonia was typically classified as community-acquired, hospital-acquired, or ventilator-associated. Over the last decade or two, however, patients presenting to the hospital have often been found to be infected with multidrug-resistant (MDR) pathogens previously associated with hospital-acquired pneumonia. Factors responsible for this phenomenon include the development and widespread use of potent oral antibiotics, earlier transfer of patients out of acute-care hospitals to their homes or various lower-acuity facilities, increased use of outpatient IV antibiotic therapy, general aging of the population, and more extensive immunomodulatory therapies. The potential involvement of these MDR pathogens has led to a revised classification system in which infection is categorized as either community-acquired pneumonia (CAP) or health care–associated pneumonia (HCAP), with subcategories of HCAP including hospital-acquired pneumonia (HAP) and ventilator-associated pneumonia (VAP). The conditions associated with HCAP and the likely pathogens are listed in Table 251-1.
Table 251-1 Clinical Conditions Associated with and Likely Pathogens in Health Care–Associated Pneumonia
Although the new classification system has been helpful in designing empirical antibiotic strategies, it is not without disadvantages. For instance, not all MDR pathogens are associated with all risk factors (Table 251-1). Therefore, this system represents a distillation of multiple risk factors, and each patient must be considered individually. For example, the risk of infection with MDR pathogens for a nursing home resident with dementia who can independently dress, ambulate, and eat is quite different from the risk for a patient who is in a chronic vegetative state with a tracheostomy and a percutaneous feeding tube in place. In addition, risk factors for MDR infection do not preclude the development of pneumonia caused by the usual CAP pathogens.
This chapter deals with pneumonia in patients who are not considered to be immunocompromised. Pneumonia in immunocompromised patients is discussed in other chapters, including Chaps. 82, 126, and 182.
Pneumonia results from the proliferation of microbial pathogens at the alveolar level and the host's response to those pathogens. Microorganisms gain access to the lower respiratory tract in several ways. The most common is by aspiration from the oropharynx. Small-volume aspiration occurs frequently during sleep (especially in the elderly) and in patients with decreased levels of consciousness. Many pathogens are inhaled as contaminated droplets. Rarely, pneumonia occurs via hematogenous spread (e.g., from tricuspid endocarditis) or by contiguous extension from an infected pleural or mediastinal space.
Mechanical factors are critically important in host defense. The hairs and turbinates of the nares catch larger inhaled particles before they reach the lower respiratory tract, and the branching architecture of the tracheobronchial tree traps particles on the airway lining, where mucociliary clearance and local antibacterial factors either clear or kill the potential pathogen. The gag reflex and the cough mechanism offer critical protection from aspiration. In addition, the normal flora adhering to mucosal cells of the oropharynx, whose components are remarkably constant, prevents pathogenic bacteria from binding and thereby decreases the risk of pneumonia caused by these more virulent bacteria.
When these barriers are overcome or when the microorganisms are small enough to be inhaled to the alveolar level, resident alveolar macrophages are extremely efficient at clearing and killing pathogens. Macrophages are assisted by local proteins (e.g., surfactant proteins A and D) that have intrinsic opsonizing properties or antibacterial or antiviral activity. Once engulfed, the pathogens—even if they are not killed by macrophages—are eliminated via either the mucociliary elevator or the lymphatics and no longer represent an infectious challenge. Only when the capacity of the alveolar macrophages to ingest or kill the microorganisms is exceeded does clinical pneumonia become manifest. In that situation, the alveolar macrophages initiate the inflammatory response to bolster lower respiratory tract defenses. The host inflammatory response, rather than the proliferation of microorganisms, triggers the clinical syndrome of pneumonia. The release of inflammatory mediators, such as interleukin (IL) 1 and tumor necrosis factor (TNF), results in fever. Chemokines, such as IL-8 and granulocyte colony-stimulating factor, stimulate the release of neutrophils and their attraction to the lung, producing both peripheral leukocytosis and increased purulent secretions. Inflammatory mediators released by macrophages and the newly recruited neutrophils create an alveolar capillary leak equivalent to that seen in the acute respiratory distress syndrome (ARDS), although in pneumonia this leak is localized (at least initially). Even erythrocytes can cross the alveolar-capillary membrane, with consequent hemoptysis. The capillary leak results in a radiographic infiltrate and rales detectable on auscultation, and hypoxemia results from alveolar filling. Moreover, some bacterial pathogens appear to interfere with the hypoxic vasoconstriction that would normally occur with fluid-filled alveoli, and this interference can result in severe hypoxemia. Increased respiratory drive in the systemic inflammatory response syndrome (SIRS) leads to respiratory alkalosis. Decreased compliance due to capillary leak, hypoxemia, increased respiratory drive, increased secretions, and occasionally infection-related bronchospasm all lead to dyspnea. If severe enough, the changes in lung mechanics secondary to reductions in lung volume and compliance and the intrapulmonary shunting of blood may cause the patient's death.
Classic pneumonia evolves through a series of pathologic changes. The initial phase is one of edema, with the presence of a proteinaceous exudate—and often of bacteria—in the alveoli. This phase is rarely evident in clinical or autopsy specimens because it is so rapidly followed by a red hepatization phase. The presence of erythrocytes in the cellular intraalveolar exudate gives this second stage its name, but neutrophils are also present and are important from the standpoint of host defense. Bacteria are occasionally seen in cultures of alveolar specimens collected during this phase. In the third phase, gray hepatization, no new erythrocytes are extravasating, and those already present have been lysed and degraded. The neutrophil is the predominant cell, fibrin deposition is abundant, and bacteria have disappeared. This phase corresponds with successful containment of the infection and improvement in gas exchange. In the final phase, resolution, the macrophage is the dominant cell type in the alveolar space, and the debris of neutrophils, bacteria, and fibrin has been cleared, as has the inflammatory response.
This pattern has been described best for pneumococcal pneumonia and may not apply to pneumonias of all etiologies, especially viral or Pneumocystis pneumonia. In VAP, respiratory bronchiolitis may precede the development of a radiologically apparent infiltrate. Because of the microaspiration mechanism, a bronchopneumonia pattern is most common in nosocomial pneumonias, whereas a lobar pattern is more common in bacterial CAP. Despite the radiographic appearance, viral and Pneumocystis pneumonias represent alveolar rather than interstitial processes.
The extensive list of potential etiologic agents in CAP includes bacteria, fungi, viruses, and protozoa. Newly identified pathogens include hantaviruses, metapneumoviruses, the coronavirus responsible for the severe acute respiratory syndrome (SARS), and community-acquired strains of methicillin-resistant Staphylococcus aureus (MRSA). Most cases of CAP, however, are caused by relatively few pathogens (Table 251-2). Although Streptococcus pneumoniae is most common, other organisms must also be considered in light of the patient's risk factors and severity of illness. In most cases, it is most useful to think of the potential causes as either "typical" bacterial pathogens or "atypical" organisms. The former category includes S. pneumoniae, Haemophilus influenzae, and (in selected patients) S. aureus and gram-negative bacilli such as Klebsiella pneumoniae and Pseudomonas aeruginosa. The "atypical" organisms include Mycoplasma pneumoniae, Chlamydophila pneumoniae, and Legionella spp. as well as respiratory viruses such as influenza viruses, adenoviruses, and respiratory syncytial viruses (RSVs). Data suggest that a virus may be responsible in up to 18% of cases of CAP that require admission to the hospital. The atypical organisms cannot be cultured on standard media, nor can they be seen on Gram's stain. The frequency and importance of atypical pathogens such as M. pneumoniae and C. pneumoniae in outpatients and Legionella in inpatients have significant implications for therapy. These organisms are intrinsically resistant to all -lactam agents and must be treated with a macrolide, a fluoroquinolone, or a tetracycline. In the ~10–15% of CAP cases that are polymicrobial, the etiology often includes a combination of typical and atypical pathogens.
Table 251-2 Microbial Causes of Community-Acquired Pneumonia, by Site of Care
Note: Pathogens are listed in descending order of frequency. ICU, intensive care unit.
aInfluenza A and B viruses, adenoviruses, respiratory syncytial viruses, parainfluenza viruses.
Anaerobes play a significant role only when an episode of aspiration has occurred days to weeks before presentation for pneumonia. The combination of an unprotected airway (e.g., in patients with alcohol or drug overdose or a seizure disorder) and significant gingivitis constitutes the major risk factor. Anaerobic pneumonias are often complicated by abscess formation and significant empyemas or parapneumonic effusions.
S. aureus pneumonia is well known to complicate influenza infection. Recently, however, MRSA strains have been reported as primary causes of CAP. While this entity is still relatively uncommon, clinicians must be aware of its potentially serious consequences, such as necrotizing pneumonia. Two important developments have led to this problem: the spread of MRSA from the hospital setting to the community and the emergence of genetically distinct strains of MRSA in the community. These novel community-acquired MRSA (CA-MRSA) strains have infected healthy individuals who have had no association with health care.
Unfortunately, despite a careful history and physical examination as well as routine radiographic studies, it is usually impossible to predict the pathogen in a case of CAP with any degree of certainty; in more than half of cases, a specific etiology is never determined. Nevertheless, it is important to consider epidemiologic and risk factors that might suggest certain pathogens (Table 251-3).
Table 251-3 Epidemiologic Factors Suggesting Possible Causes of Community-Acquired Pneumonia
In the United States, ~80% of the 4 million CAP cases that occur annually are treated on an outpatient basis, and ~20% are treated in the hospital. CAP results in more than 600,000 hospitalizations, 64 million days of restricted activity, and 45,000 deaths annually. The overall yearly cost associated with CAP is estimated at $9–10 billion (U.S.). The incidence rates are highest at the extremes of age. Although the overall annual figure in the United States is 12 cases per 1000 persons, the figure is 12–18 per 1000 among children <4 years of age and 20 per 1000 among persons >60 years of age.
The risk factors for CAP in general and for pneumococcal pneumonia in particular have implications for treatment regimens. Risk factors for CAP include alcoholism, asthma, immunosuppression, institutionalization, and an age of 70 years versus 60–69 years. Risk factors for pneumococcal pneumonia include dementia, seizure disorders, heart failure, cerebrovascular disease, alcoholism, tobacco smoking, chronic obstructive pulmonary disease, and HIV infection. CA-MRSA infection is more likely in Native Americans, homeless youths, men who have sex with men, prison inmates, military recruits, children in day-care centers, and athletes such as wrestlers. The Enterobacteriaceae tend to affect patients who have recently been hospitalized and/or received antibiotic therapy or who have comorbidities such as alcoholism, heart failure, or renal failure. P. aeruginosa may also infect these patients as well as those with severe structural lung disease. Risk factors for Legionella infection include diabetes, hematologic malignancy, cancer, severe renal disease, HIV infection, smoking, male gender, and a recent hotel stay or ship cruise. (Many of these risk factors would now reclassify as HCAP some cases that were previously designated CAP.)
CAP can vary from indolent to fulminant in presentation and from mild to fatal in severity. The various signs and symptoms, which depend on the progression and severity of the infection, include both constitutional findings and manifestations limited to the lung and its associated structures. In light of the pathobiology of the disease, many of the findings are to be expected.
The patient is frequently febrile, with a tachycardic response, and may have chills and/or sweats and cough that is either nonproductive or productive of mucoid, purulent, or blood-tinged sputum. In accordance with the severity of infection, the patient may be able to speak in full sentences or may be very short of breath. If the pleura is involved, the patient may experience pleuritic chest pain. Up to 20% of patients may have gastrointestinal symptoms such as nausea, vomiting, and/or diarrhea. Other symptoms may include fatigue, headache, myalgias, and arthralgias.
Findings on physical examination vary with the degree of pulmonary consolidation and the presence or absence of a significant pleural effusion. An increased respiratory rate and use of accessory muscles of respiration are common. Palpation may reveal increased or decreased tactile fremitus, and the percussion note can vary from dull to flat, reflecting underlying consolidated lung and pleural fluid, respectively. Crackles, bronchial breath sounds, and possibly a pleural friction rub may be heard on auscultation. The clinical presentation may not be so obvious in the elderly, who may initially display new-onset or worsening confusion and few other manifestations. Severely ill patients who have septic shock secondary to CAP are hypotensive and may have evidence of organ failure.
When confronted with possible CAP, the physician must ask two questions: Is this pneumonia, and, if so, what is the etiology? The former question is typically answered by clinical and radiographic methods, whereas the latter requires the aid of laboratory techniques.
The differential diagnosis includes both infectious and noninfectious entities such as acute bronchitis, acute exacerbations of chronic bronchitis, heart failure, pulmonary embolism, and radiation pneumonitis. The importance of a careful history cannot be overemphasized. For example, known cardiac disease may suggest worsening pulmonary edema, while underlying carcinoma may suggest lung injury secondary to radiation. Epidemiologic clues, such as recent travel to areas with known endemic pathogens, may alert the physician to specific possibilities (Table 251-3).
Unfortunately, the sensitivity and specificity of the findings on physical examination are less than ideal, averaging 58% and 67%, respectively. Therefore, chest radiography is often necessary to help differentiate CAP from other conditions. Radiographic findings serve as a baseline and may include risk factors for increased severity (e.g., cavitation or multilobar involvement). Occasionally, radiographic results suggest an etiologic diagnosis. For example, pneumatoceles suggest infection with S. aureus, and an upper-lobe cavitating lesion suggests tuberculosis. CT is rarely necessary but may be of value in a patient with suspected postobstructive pneumonia caused by a tumor or foreign body. For cases managed on an outpatient basis, the clinical and radiologic assessment is usually all that is done before treatment is started since most laboratory test results are not available soon enough to influence initial management. In certain cases, however (e.g., influenza virus infection), the availability of rapid point-of-care diagnostic tests and access to specific drugs for treatment and prevention can be very important.
The etiology of pneumonia usually cannot be determined on the basis of clinical presentation; instead, the physician must rely upon the laboratory for support. Except for the 2% of CAP patients who are admitted to the intensive care unit (ICU), no data exist to show that treatment directed at a specific pathogen is statistically superior to empirical therapy. The benefits of establishing a microbial etiology can therefore be questioned, particularly in light of the cost of diagnostic testing. However, a number of reasons can be advanced for attempting an etiologic diagnosis. Identification of an unexpected pathogen allows narrowing of the initial empirical regimen, which decreases antibiotic selection pressure and may lessen the risk of resistance. Pathogens with important public safety implications, such as Mycobacterium tuberculosis and influenza virus, may be found in some cases. Finally, without culture and susceptibility data, trends in resistance cannot be followed accurately, and appropriate empirical therapeutic regimens are harder to devise.
Gram's Stain and Culture of Sputum
The main purpose of the sputum Gram's stain is to ensure that a sample is suitable for culture. However, Gram's staining may also help to identify certain pathogens (e.g., S. pneumoniae, S. aureus, and gram-negative bacteria) by their characteristic appearance. To be adequate for culture, a sputum sample must have >25 neutrophils and <10 squamous epithelial cells per low-power field. The sensitivity and specificity of the sputum Gram's stain and culture are highly variable; even in cases of proven bacteremic pneumococcal pneumonia, the yield of positive cultures from sputum samples is 50%.
Some patients, particularly elderly individuals, may not be able to produce an appropriate expectorated sputum sample. Others may already have started a course of antibiotics, which can interfere with results, at the time a sample is obtained. The inability to produce sputum can be a consequence of dehydration, and the correction of this condition may result in increased sputum production and a more obvious infiltrate on chest radiography. For patients admitted to the ICU and intubated, a deep-suction aspirate or bronchoalveolar lavage sample should be sent to the microbiology laboratory as soon as possible. Since the etiologies in severe CAP are somewhat different from those in milder disease (Table 251-2), the greatest benefit of staining and culturing respiratory secretions is to alert the physician of unsuspected and/or resistant pathogens and to permit appropriate modification of therapy. Other stains and cultures may be useful as well. For suspected tuberculosis or fungal infection, specific stains are available. Cultures of pleural fluid obtained from effusions >1 cm in height on a lateral decubitus chest radiograph may also be helpful.
The yield from blood cultures, even those obtained before antibiotic therapy, is disappointingly low. Only ~5–14% of cultures of blood from patients hospitalized with CAP are positive, and the most frequently isolated pathogen is S. pneumoniae. Since recommended empirical regimens all provide pneumococcal coverage, a blood culture positive for this pathogen has little, if any, effect on clinical outcome. However, susceptibility data may allow a switch from a broader-spectrum regimen (e.g., a fluoroquinolone or -lactam plus a macrolide) to penicillin in appropriate cases. Because of the low yield and the lack of significant impact on outcome, blood cultures are no longer considered de rigueur for all hospitalized CAP patients. Certain high-risk patients—including those with neutropenia secondary to pneumonia, asplenia, or complement deficiencies; chronic liver disease; or severe CAP—should have blood cultured.
Two commercially available tests detect pneumococcal and certain Legionella antigens in urine. The test for Legionella pneumophila detects only serogroup 1, but this serogroup accounts for most community-acquired cases of Legionnaires' disease. The sensitivity and specificity of the Legionella urine antigen test are as high as 90% and 99%, respectively. The pneumococcal urine antigen test is also quite sensitive and specific (80% and >90%, respectively). Although false-positive results can be obtained with samples from colonized children, the test is generally reliable. Both tests can detect antigen even after the initiation of appropriate antibiotic therapy and after weeks of illness. Other antigen tests include a rapid test for influenza virus and direct fluorescent antibody tests for influenza virus and RSV, although the test for RSV is only poorly sensitive.
Polymerase Chain Reaction
Polymerase chain reaction (PCR) tests are available for a number of pathogens, including L. pneumophila and mycobacteria. In addition, a multiplex PCR can detect the nucleic acid of Legionella spp., M. pneumoniae, and C. pneumoniae. However, the use of these PCR assays is generally limited to research studies.
A fourfold rise in specific IgM antibody titer between acute- and convalescent-phase serum samples is generally considered diagnostic of infection with the pathogen in question. In the past, serologic tests were used to help identify atypical pathogens as well as some typical but relatively unusual organisms, such as Coxiella burnetii. Recently, however, they have fallen out of favor because of the time required to obtain a final result for the convalescent-phase sample.
Community-Acquired Pneumonia: Treatment
Site of Care
The decision to hospitalize a patient with CAP must take into consideration diminishing health care resources and rising costs of treatment. The cost of inpatient management exceeds that of outpatient treatment by a factor of 20 and accounts for most CAP-related expenditures. Certain patients clearly can be managed at home, and others clearly require treatment in the hospital, but the choice is sometimes difficult. Tools that objectively assess the risk of adverse outcomes, including severe illness and death, may minimize unnecessary hospital admissions and help to identify patients who will benefit from hospital care. There are currently two sets of criteria: the Pneumonia Severity Index (PSI), a prognostic model used to identify patients at low risk of dying; and the CURB-65 criteria, a severity-of-illness score.
To determine the PSI, points are given for 20 variables, including age, coexisting illness, and abnormal physical and laboratory findings. On the basis of the resulting score, patients are assigned to one of five classes with the following mortality rates: class 1, 0.1%; class 2, 0.6%; class 3, 2.8%; class 4, 8.2%; and class 5, 29.2%. Clinical trials have demonstrated that routine use of the PSI results in lower admission rates for class 1 and class 2 patients. Patients in classes 4 and 5 should be admitted to the hospital, while those in class 3 should ideally be admitted to an observation unit until a further decision can be made.
The CURB-65 criteria include five variables: confusion (C); urea >7 mmol/L (U); respiratory rate 30/min (R); blood pressure, systolic 90 mmHg or diastolic 60 mmHg (B); and age 65 years (65). Patients with a score of 0, among whom the 30-day mortality rate is 1.5%, can be treated outside the hospital. With a score of 2, the 30-day mortality rate is 9.2%, and patients should be admitted to the hospital. Among patients with scores of 3, mortality rates are 22% overall; these patients may require admission to an ICU.
At present, it is difficult to say which assessment tool is superior. The PSI is less practical in a busy emergency-room setting because of the need to assess 20 variables. While the CURB-65 criteria are easily remembered, they have not been studied as extensively. Whichever system is used, these objective criteria must always be tempered by careful consideration of factors relevant to individual patients, including the ability to comply reliably with an oral antibiotic regimen and the resources available to the patient outside the hospital.
Antimicrobial resistance is a significant problem that threatens to diminish our therapeutic armamentarium. Misuse of antibiotics results in increased antibiotic selection pressure that can affect resistance locally or even globally by clonal dissemination. For CAP, the main resistance issues currently involve S. pneumoniae and CA-MRSA.
In general, pneumococcal resistance is acquired (1) by direct DNA incorporation and remodeling resulting from contact with closely related oral commensal bacteria, (2) by the process of natural transformation, or (3) by mutation of certain genes.
Pneumococcal strains are classified as sensitive to penicillin if the minimal inhibitory concentration (MIC) is 0.06 g/mL, as intermediate if the MIC is 0.1–1.0 g/mL, and as resistant if the MIC is 2 g/mL. Strains resistant to drugs from three or more antimicrobial classes with different mechanisms of action are considered MDR isolates. Pneumococcal resistance to -lactam drugs is due solely to the presence of low-affinity penicillin-binding proteins. The propensity for pneumococcal resistance to penicillin to be associated with reduced susceptibility to other drugs, such as macrolides, tetracyclines, and trimethoprim-sulfamethoxazole (TMP-SMX), is of concern. In the United States, 58.9% of penicillin-resistant pneumococcal isolates from blood cultures are also resistant to macrolides. Penicillin is an appropriate agent for the treatment of pneumococcal infection caused by strains with MICs of 1 g/mL. For infections caused by pneumococcal strains with penicillin MICs of 2–4 g/mL, the data are conflicting; some studies suggest no increase in treatment failure with penicillin, while others suggest increased rates of death or complications. For strains of S. pneumoniae with intermediate levels of resistance, higher doses of the drug should be used. Risk factors for drug-resistant pneumococcal infection include recent antimicrobial therapy, an age of <2 years or >65 years, attendance at day-care centers, recent hospitalization, and HIV infection. Fortunately, resistance to penicillin appears to be reaching a plateau.
In contrast, resistance to macrolides is increasing through several mechanisms, including target-site modification and the presence of an efflux pump. Target-site modification is caused by ribosomal methylation in 23S rRNA encoded by the ermB gene and results in resistance to macrolides, lincosamides, and streptogramin B–type antibiotics. This MLSB phenotype is associated with high-level resistance, with typical MICs of 64 g/mL. The efflux mechanism encoded by the mef gene (M phenotype) is usually associated with low-level resistance (MICs, 1–32 g/mL). These two mechanisms account for ~45% and ~65%, respectively, of resistant pneumococcal isolates in the United States. Some pneumococcal isolates with both the erm and mef genes have been identified, but the exact significance of this finding is unknown. High-level resistance to macrolides is more common in Europe, whereas lower-level resistance seems to predominate in North America. Although clinical failures with macrolides have been reported, many experts think that these drugs still have a role to play in the management of pneumococcal pneumonia in North America.
Pneumococcal resistance to fluoroquinolones (e.g., ciprofloxacin and levofloxacin) has been reported. Changes can occur in one or both target sites (topoisomerases II and IV); changes in these two sites usually result from mutations in the gyrA and parC genes, respectively. The increasing number of pneumococcal isolates that, although susceptible to fluoroquinolones, already have a mutation in one target site is of concern. Such organisms may be more likely to undergo a second step mutation that will render them fully resistant to fluoroquinolones. In addition, an efflux pump may play a role in pneumococcal resistance to fluoroquinolones.
CAP due to MRSA may be caused by infection with the classic hospital-acquired strains or with the more recently identified, genotypically and phenotypically distinct community-acquired strains. Most infections with the former strains have been acquired either directly or indirectly by contact with the health care environment and, although classified as HAP in the past, would now be classified as HCAP. In some hospitals, CA-MRSA strains are displacing the classic hospital-acquired strains—a trend suggesting that the newer strains may be more robust.
Methicillin resistance in S. aureus is determined by the mecA gene, which encodes for resistance to all -lactam drugs. At least five staphylococcal chromosomal cassette mec (SCCmec) types have been described. The typical hospital-acquired strain usually has type II or III, whereas CA-MRSA has a type IV SCCmec element. CA-MRSA isolates tend to be less resistant than the older hospital-acquired strains and are often susceptible to TMP-SMX, clindamycin, and tetracycline in addition to vancomycin and linezolid. However, CA-MRSA strains may also carry genes for superantigens, such as enterotoxins B and C and Panton-Valentine leukocidin, a membrane-tropic toxin that can create cytolytic pores in polymorphonuclear neutrophils, monocytes, and macrophages.
A detailed discussion of resistance among gram-negative bacilli is beyond the scope of this chapter (see Chap. 143). Fluoroquinolone resistance among isolates of Escherichia coli from the community appears to be increasing. Enterobacter spp. are typically resistant to cephalosporins; the drugs of choice for use against these bacteria are usually fluoroquinolones or carbapenems. Similarly, when infections due to bacteria producing extended-spectrum -lactamases (ESBLs) are documented or suspected, a fluoroquinolone or a carbapenem should be used; these MDR strains are more likely to be involved in HCAP.
Initial Antibiotic Management
Since the physician rarely knows the etiology of CAP at the outset of treatment, initial therapy is usually empirical and is designed to cover the most likely pathogens (Table 251-4). In all cases, antibiotic treatment should be initiated as expeditiously as possible.
Table 251-4 Empirical Antibiotic Treatment of Community-Acquired Pneumonia
Previously healthy and no antibiotics in past 3 months
A macrolide [clarithromycin (500 mg PO bid) or azithromycin (500 mg PO once, then 250 mg od)] or
Doxycycline (100 mg PO bid)
Comorbidities or antibiotics in past 3 months: select an alternative from a different class
A respiratory fluoroquinolone [moxifloxacin (400 mg PO od), gemifloxacin (320 mg PO od), levofloxacin (750 mg PO od)] or
A -lactam [preferred: high-dose amoxicillin (1 g tid) or amoxicillin/clavulanate (2 g bid); alternatives: ceftriaxone (1–2 g IV od), cefpodoxime (200 mg PO bid), cefuroxime (500 mg PO bid)] plus a macrolidea
In regions with a high rate of "high-level" pneumococcal macrolide resistance,b consider alternatives listed above for patients with comorbidities.
A respiratory fluoroquinolone [moxifloxacin (400 mg PO or IV od), gemifloxacin (320 mg PO od), levofloxacin (750 mg PO or IV od)]
A -lactamc [cefotaxime (1–2 g IV q8h), ceftriaxone (1–2 g IV od), ampicillin (1–2 g IV q4–6h), ertapenem (1 g IV od in selected patients)] plus a macrolided [oral clarithromycin or azithromycin (as listed above for previously healthy patients) or IV azithromycin (1 g once, then 500 mg od)]
A -lactame [cefotaxime (1–2 g IV q8h), ceftriaxone (2 g IV od), ampicillin-sulbactam (2 g IV q8h)] plus
Azithromycin or a fluoroquinolone (as listed above for inpatients, non-ICU)
If Pseudomonas is a consideration
An antipneumococcal, antipseudomonal -lactam [piperacillin/tazobactam (4.5 g IV q6h), cefepime (1–2 g IV q12h), imipenem (500 mg IV q6h), meropenem (1 g IV q8h)] plus either ciprofloxacin (400 mg IV q12h) or levofloxacin (750 mg IV od)
The above -lactams plus an aminoglycoside [amikacin (15 mg/kg od) or tobramycin (1.7 mg/kg od) and azithromycin]
The above -lactamsfplus an aminoglycoside plus an antipneumococcal fluoroquinolone
If CA-MRSA is a consideration
Add linezolid (600 mg IV q12h) or vancomycin (1 g IV q12h)
Note: CA-MRSA, community-acquired methicillin-resistant Staphylococcus aureus; ICU, intensive care unit.
aDoxycycline (100 mg PO bid) is an alternative to the macrolide.
bMICs of >16 g/mL in 25% of isolates.
cA respiratory fluoroquinolone should be used for penicillin-allergic patients.
dDoxycycline (100 mg IV q12h) is an alternative to the macrolide.
eFor penicillin-allergic patients, use a respiratory fluoroquinolone and aztreonam (2 g IV q8h).
The CAP treatment guidelines in the United States (summarized in Table 251-4) represent joint statements from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS); the Canadian guidelines come from the Canadian Infectious Disease Society and the Canadian Thoracic Society. In these guidelines, coverage is always provided for the pneumococcus and the atypical pathogens. In contrast, guidelines from some European countries do not always include atypical coverage based on local epidemiologic data. The U.S.-Canadian approach is supported by retrospective data from almost 13,000 patients >65 years of age. Atypical pathogen coverage provided by a macrolide or a fluoroquinolone has been associated with a significant reduction in mortality rates compared with those for -lactam coverage alone.
Therapy with a macrolide or a fluoroquinolone within the previous 3 months is associated with an increased likelihood of infection with a macrolide- or fluoroquinolone-resistant strain of S. pneumoniae. For this reason, a fluoroquinolone-based regimen should be used for patients recently given a macrolide, and vice versa (Table 251-4). Telithromycin, a ketolide derived from the macrolide class, differs from the macrolides in that it binds to bacteria more avidly and at two sites rather than one. This drug is active against pneumococci resistant to penicillins, macrolides, and fluoroquinolones. Its future role in the outpatient management of CAP will depend on the evaluation of its safety by the U.S. Food and Drug Administration.
Once the etiologic agent(s) and susceptibilities are known, therapy may be altered to target the specific pathogen(s). However, this decision is not always straightforward. If blood cultures yield S. pneumoniae sensitive to penicillin after 2 days of treatment with a macrolide plus a -lactam or a fluoroquinolone, should therapy be switched to penicillin? Penicillin alone would not be effective in the potential 15% of cases with atypical co-infection. No standard approach exists. Some experts would argue that pneumococcal coverage by a switch to penicillin is appropriate, while others would opt for continued coverage of both the pneumococcus and atypical pathogens. One compromise would be to continue atypical coverage with either a macrolide or a fluoroquinolone for a few more days and then to complete the treatment course with penicillin alone. In all cases, the individual patient and the various risk factors must be considered.
Management of bacteremic pneumococcal pneumonia is also controversial. Data from nonrandomized studies suggest that combination therapy (e.g., with a macrolide and a -lactam) is associated with a lower mortality rate than monotherapy, particularly in severely ill patients. The exact reason is unknown, but explanations include possible atypical co-infection or the immunomodulatory effects of the macrolides.
For patients with CAP who are admitted to the ICU, the risk of infection with P. aeruginosa or CA-MRSA is increased, and coverage should be considered when a patient has risk factors or a Gram's stain suggestive of these pathogens (Table 251-4). The main risk factors for P. aeruginosa infection are structural lung disease (e.g., bronchiectasis) and recent treatment with antibiotics or glucocorticoids. If CA-MRSA infection is suspected, either linezolid or vancomycin should be added to the initial empirical regimen.
Although hospitalized patients have traditionally received initial therapy by the IV route, some drugs—particularly the fluoroquinolones—are very well absorbed and can be given orally from the outset to select patients. For patients initially treated IV, a switch to oral treatment is appropriate as long as the patient can ingest and absorb the drugs, is hemodynamically stable, and is showing clinical improvement.
The duration of treatment for CAP has recently generated considerable interest. Patients have usually been treated for 10–14 days, but recent studies with fluoroquinolones and telithromycin suggest that a 5-day course is sufficient for otherwise uncomplicated CAP. A longer course is required for patients with bacteremia, metastatic infection, or infection with a particularly virulent pathogen, such as P. aeruginosa or CA-MRSA. Longer-term therapy should also be considered if initial treatment was ineffective and in most cases of severe CAP. Data from studies with azithromycin, which suggest 3–5 days of treatment for outpatient-managed CAP, cannot be extrapolated to other drugs because of the extremely long half-life of azithromycin.
Patients may be discharged from the hospital once they are clinically stable and have no active medical problems requiring ongoing hospital care. The site of residence after discharge (in a nursing home, at home with family, at home alone) is an important consideration, particularly for elderly patients.
In addition to appropriate antimicrobial therapy, certain general considerations apply in dealing with either CAP or HAP. Adequate hydration, oxygen therapy for hypoxemia, and assisted ventilation when necessary are critical to the success of therapy. Patients with severe CAP who remain hypotensive despite fluid resuscitation may have adrenal insufficiency and may respond to glucocorticoid treatment. Immunomodulatory therapy in the form of drotrecogin alfa (activated) should be considered for CAP patients with persistent septic shock and APACHE II scores of 25, particularly if the infection is caused by S. pneumoniae.
Failure to Improve
Patients who are slow to respond to therapy should be reevaluated at about day 3 (sooner if their condition is worsening rather than simply not improving), and a number of possible scenarios should be considered. (1) Is this a noninfectious condition? (2) If this is an infection, is the correct pathogen being targeted? (3) Is this a superinfection with a new nosocomial pathogen? A number of noninfectious conditions can mimic pneumonia, including pulmonary edema, pulmonary embolism, lung carcinoma, radiation and hypersensitivity pneumonitis, and connective tissue disease involving the lungs. If the patient has CAP and treatment is aimed at the correct pathogen, the lack of response may be explained in a number of ways. The pathogen may be resistant to the drug selected, or a sequestered focus (e.g., a lung abscess or empyema) may be blocking access of the antibiotic(s) to the pathogen. Alternatively, the patient may be getting either the wrong drug or the correct drug at the wrong dose or frequency of administration. It is also possible that CAP is the correct diagnosis but that a different pathogen (e.g., M. tuberculosis or a fungus) is the cause. In addition, nosocomial superinfections—both pulmonary and extrapulmonary—are possible explanations for persistence. In all cases of delayed response or deteriorating condition, the patient must be carefully reassessed and appropriate studies initiated. These studies may include such diverse procedures as CT and bronchoscopy.
As in other severe infections, common complications of severe CAP include respiratory failure, shock and multiorgan failure, bleeding diatheses, and exacerbation of comorbid illnesses. Three particularly noteworthy conditions are metastatic infection, lung abscess, and complicated pleural effusion. Metastatic infection (e.g., brain abscess or endocarditis), although unusual, deserves immediate attention by the physician, with a detailed workup and proper treatment. Lung abscess may occur in association with aspiration or with infection caused by a single CAP pathogen, such CA-MRSA, P. aeruginosa, or (rarely) S. pneumoniae. Aspiration pneumonia is typically a mixed polymicrobial infection involving both aerobes and anaerobes. In either scenario, drainage should be established, and antibiotics that cover the known or suspected pathogens should be administered. A significant pleural effusion should be tapped for both diagnostic and therapeutic purposes. If the fluid has a pH of <7, a glucose level of <2.2 mmol/L, and a lactate dehydrogenase concentration of >1000 U/L or if bacteria are seen or cultured, then the fluid should be drained; a chest tube is usually required.
Fever and leukocytosis usually resolve within 2 and 4 days, respectively, in otherwise healthy patients with CAP, but physical findings may persist longer. Chest radiographic abnormalities are slowest to resolve and may require 4–12 weeks to clear, with the speed of clearance depending on the patient's age and underlying lung disease. For a patient whose condition is improving and who (if hospitalized) has been discharged, a follow-up radiograph can be done ~4–6 weeks later. If relapse or recurrence is documented, particularly in the same lung segment, the possibility of an underlying neoplasm must be considered.
The prognosis of CAP depends on the patient's age, comorbidities, and site of treatment (inpatient or outpatient). Young patients without comorbidity do well and usually recover fully after ~2 weeks. Older patients and those with comorbid conditions can take several weeks longer to recover fully. The overall mortality rate for the outpatient group is <1%. For patients requiring hospitalization, the overall mortality rate is estimated at 10%, with ~50% of the deaths directly attributable to pneumonia.
The main preventive measure is vaccination. The recommendations of the Advisory Committee on Immunization Practices should be followed for influenza and pneumococcal vaccines. In the event of an influenza outbreak, unprotected patients at risk from complications should be vaccinated immediately and given chemoprophylaxis with either oseltamivir or zanamivir for 2 weeks—i.e., until vaccine-induced antibody levels are sufficiently high. Because of an increased risk of pneumococcal infection, even among patients without obstructive lung disease, smokers should be strongly encouraged to stop smoking.
Most research on VAP has focused on illness in the hospital setting. However, the information and principles based on this research can be applied to HCAP not associated with ventilator use as well. The main rationale for the new designation HCAP is that the pathogens and treatment strategies for VAP are more similar to those for HAP than to those for pure CAP. The greatest difference between VAP and HCAP/HAP—and the greatest similarity of VAP to CAP—is the return to dependence on expectorated sputum for a microbiologic diagnosis, which is further complicated by the frequent colonization with pathogens among patients in the hospital or other health care–associated settings.
Potential etiologic agents of VAP include both MDR and non-MDR bacterial pathogens (Table 251-5). The non-MDR group is nearly identical to the pathogens found in severe CAP (Table 251-2); it is not surprising that such pathogens predominate if VAP develops in the first 5–7 days of the hospital stay. However, if patients have other risk factors for HCAP, MDR pathogens are a consideration, even early in the hospital course. The relative frequency of individual MDR pathogens can vary significantly from hospital to hospital and even between different critical care units within the same institution. Many hospitals have problems with P. aeruginosa and MRSA, but other MDR pathogens are often institution-specific.
Table 251-5 Microbiologic Causes of Ventilator-Associated Pneumonia
Less commonly, fungal and viral pathogens cause VAP, most frequently affecting severely immunocompromised patients. Rarely, community-associated viruses cause miniepidemics, usually when introduced by ill health care workers.
Pneumonia is a common complication among patients requiring mechanical ventilation. Prevalence estimates vary between 6 and 52 cases per 100 patients, depending on the population studied. On any given day in the ICU, an average of 10% of patients will have pneumonia—VAP in the overwhelming majority of cases. The frequency of diagnosis is not static but changes with the duration of mechanical ventilation, with the highest hazard ratio in the first 5 days and a plateau in additional cases (1% per day) after ~2 weeks. However, the cumulative rate among patients who remain ventilated for as long as 30 days is as high as 70%. These rates often do not reflect the recurrence of VAP in the same patient. Once a ventilated patient is transferred to a chronic care facility or to home, the incidence of pneumonia drops significantly, especially in the absence of other risk factors for pneumonia.
Three factors are critical in the pathogenesis of VAP: colonization of the oropharynx with pathogenic microorganisms, aspiration of these organisms from the oropharynx into the lower respiratory tract, and compromise of the normal host defense mechanisms. Most risk factors and their corresponding prevention strategies pertain to one of these three factors (Table 251-6).
Table 251-6 Pathogenic Mechanisms and Corresponding Prevention Strategies for Ventilator-Associated Pneumonia
Oropharyngeal colonization with pathogenic bacteria
Elimination of normal flora
Avoidance of prolonged antibiotic courses
Large-volume oropharyngeal aspiration around time of intubation
Short course of prophylactic antibiotics for comatose patientsa
Postpyloric enteral feedingb; avoidance of high gastric residuals, prokinetic agents
Bacterial overgrowth of stomach
Avoidance of gastrointestinal bleeding due to prophylactic agents that raise gastric pHb; selective decontamination of digestive tract with nonabsorbable antibioticsb
Cross-infection from other colonized patients
Hand washing, especially with alcohol-based hand rub; intensive infection control educationa; isolation; proper cleaning of reusable equipment
Endotracheal intubation; avoidance of sedation; decompression of small-bowel obstruction
Microaspiration around endotracheal tube
Prolonged duration of ventilation
Daily awakening from sedation,a weaning protocolsa
Abnormal swallowing function
Early percutaneous tracheostomya
Secretions pooled above endotracheal tube
Head of bed elevateda; continuous aspiration of subglottic secretions with specialized endotracheal tubea; avoidance of reintubation; minimization of sedation and patient transport
Altered lower respiratory host defenses
Tight glycemic controla; lowering of hemoglobin transfusion threshold; specialized enteral feeding formula
aStrategies demonstrated to be effective in at least one randomized controlled trial.
bStrategies with negative randomized trials or conflicting results.
The most obvious risk factor is the endotracheal tube (ET), which bypasses the normal mechanical factors preventing aspiration. While the presence of an ET may prevent large-volume aspiration, microaspiration is actually enhanced by secretions pooling above the cuff. The ET and the concomitant need for suctioning can damage the tracheal mucosa, thereby facilitating tracheal colonization. In addition, pathogenic bacteria can form a glycocalyx biofilm on the ET surface that protects them from both antibiotics and host defenses. The bacteria can also be dislodged during suctioning and can reinoculate the trachea, or tiny fragments of glycocalyx can embolize to distal airways, carrying bacteria with them.
In a high percentage of critically ill patients, the normal oropharyngeal flora is replaced by pathogenic microorganisms. The most important risk factors are antibiotic selection pressure, cross-infection from other infected/colonized patients or contaminated equipment, and malnutrition.
How the lower respiratory tract defenses become overwhelmed remains poorly understood. Almost all intubated patients experience microaspiration and are at least transiently colonized with pathogenic bacteria. However, only around one-third of colonized patients develop VAP. Severely ill patients with sepsis and trauma appear to enter a state of immunoparalysis several days after admission to the ICU—a time that corresponds to the greatest risk of developing VAP. The mechanism of this immunosuppression is not clear, although several factors have been suggested. Hyperglycemia affects neutrophil function, and recent trials suggest that keeping the blood sugar close to normal with exogenous insulin may have beneficial effects, including a decreased risk of infection. More frequent transfusions, especially of leukocyte-depleted red blood cells, also affect the immune response positively.
The clinical manifestations of VAP are generally the same as for all other forms of pneumonia: fever, leukocytosis, increase in respiratory secretions, and pulmonary consolidation on physical examination, along with a new or changing radiographic infiltrate. The frequency of abnormal chest radiographs before the onset of pneumonia in intubated patients and the limitations of portable radiographic technique make interpretation of radiographs more difficult than in patients who are not intubated. Other clinical features may include tachypnea, tachycardia, worsening oxygenation, and increased minute ventilation.
No single set of criteria is reliably diagnostic of pneumonia in a ventilated patient. The inability to identify such patients compromises efforts to prevent and treat VAP and even calls into question estimates of the impact of VAP on mortality rates.
Application of clinical criteria consistently results in overdiagnosis of VAP, largely because of three common findings in at-risk patients: (1) tracheal colonization with pathogenic bacteria in patients with ETs, (2) multiple alternative causes of radiographic infiltrates in mechanically ventilated patients, and (3) the high frequency of other sources of fever in critically ill patients. The differential diagnosis of VAP includes a number of entities, such as atypical pulmonary edema, pulmonary contusion and/or hemorrhage, hypersensitivity pneumonitis, ARDS, and pulmonary embolism. Clinical findings in ventilated patients with fever and/or leukocytosis may have alternative causes, including antibiotic-associated diarrhea, sinusitis, urinary tract infection, pancreatitis, and drug fever. Conditions mimicking pneumonia are often documented in patients in whom VAP has been ruled out by accurate diagnostic techniques. Most of these alternative diagnoses do not require antibiotic treatment; require antibiotics different from those used to treat VAP; or require some additional intervention, such as surgical drainage or catheter removal, for optimal management.
This diagnostic dilemma has led to debate and controversy. The major question is whether a quantitative-culture approach as a means of eliminating false-positive clinical diagnoses is superior to the clinical approach enhanced by principles learned from quantitative-culture studies. The recent IDSA/ATS guidelines for HCAP suggest that either approach is clinically valid.
The essence of the quantitative-culture approach is to discriminate between colonization and true infection by determining the bacterial burden. The more distal in the respiratory tree the diagnostic sampling, the more specific the results and therefore the lower the threshold of growth necessary to diagnose pneumonia and exclude colonization. For example, a quantitative endotracheal aspirate yields proximate samples, and the diagnostic threshold is 106 cfu/mL. The protected specimen brush method, in contrast, obtains distal samples and has a threshold of 103 cfu/mL. Conversely, sensitivity declines as more distal secretions are obtained, especially when they are collected blindly (i.e., by a technique other than bronchoscopy). Additional tests that may increase the diagnostic yield include Gram's stain, differential cell counts, staining for intracellular organisms, and detection of local protein levels elevated in response to infection.
Several studies have compared patient cohorts managed by the various quantitative-culture methods. While these studies documented issues of relative sensitivity and specificity, outcomes were not significantly different for the various groups of patients. The IDSA/ATS guidelines have suggested that all these methods are appropriate and that the choice depends on availability and local expertise.
The Achilles heel of the quantitative approach is the effect of antibiotic therapy. With sensitive microorganisms, a single antibiotic dose can reduce colony counts below the diagnostic threshold. Recent changes in antibiotic therapy are the most significant. After 3 days of consistent antibiotic therapy for another infection prior to suspicion of pneumonia, the accuracy of diagnostic tests for pneumonia is unaffected. Conversely, colony counts above the diagnostic threshold during antibiotic therapy suggest that the current antibiotics are ineffective. Even the normal host response may be sufficient to reduce quantitative-culture counts below the diagnostic threshold by the time of sampling. In short, expertise in quantitative-culture techniques is critical, with a specimen obtained as soon as pneumonia is suspected and before antibiotic therapy is initiated or changed.
In a study comparing the quantitative with the clinical approach, use of bronchoscopic quantitative cultures resulted in significantly less antibiotic use at 14 days after study entry and lower rates of mortality and severity-adjusted mortality at 28 days. In addition, more alternative sites of infection were found in patients randomized to the quantitative-culture strategy. A critical aspect of this study was that antibiotic treatment was initiated only in patients whose gram-stained respiratory sample was positive or who displayed signs of hemodynamic instability. Fewer than half as many patients were treated for pneumonia in the bronchoscopy group, and only one-third as many microorganisms were cultured.
The lack of specificity of a clinical diagnosis of VAP has led to efforts to improve the diagnostic criteria. The Clinical Pulmonary Infection Score (CPIS) was developed by weighting of the various clinical criteria usually used for the diagnosis of VAP (Table 251-7). Use of the CPIS allows the selection of low-risk patients who may need only short-course antibiotic therapy or no treatment at all. Moreover, studies have demonstrated that the absence of bacteria in gram-stained endotracheal aspirates makes pneumonia an unlikely cause of fever or pulmonary infiltrates. These findings, coupled with a heightened awareness of the alternative diagnoses possible in patients with suspected VAP, can prevent inappropriate treatment for this disease. Furthermore, data show that the absence of an MDR pathogen in tracheal aspirate cultures eliminates the need for MDR coverage when empirical antibiotic therapy is narrowed. Since the most likely explanations for the mortality benefit of bronchoscopic quantitative cultures are decreased antibiotic selection pressure (which reduces the risk of subsequent infection with MDR pathogens) and identification of alternative sources of infection, a clinical diagnostic approach that incorporates such principles may result in similar outcomes.
Many studies have demonstrated higher mortality rates with inappropriate than with appropriate empirical antibiotic therapy. The key to appropriate antibiotic management of VAP is an appreciation of the patterns of resistance of the most likely pathogens in any given patient.
If it were not for the risk of infection with MDR pathogens (Table 251-1), VAP could be treated with the same antibiotics used for severe CAP. However, antibiotic selection pressure leads to the frequent involvement of MDR pathogens by selecting either for drug-resistant isolates of common pathogens (MRSA and ESBL-positive Enterobacteriaceae) or for intrinsically resistant pathogens (P. aeruginosa and Acinetobacter spp.). Frequent use of -lactam drugs, especially cephalosporins, appears to be the major risk factor for infection with MRSA and ESBL-positive strains.
P. aeruginosa has demonstrated the ability to develop resistance to all routinely used antibiotics. Unfortunately, even if initially sensitive, P. aeruginosa isolates have also shown a propensity to develop resistance during treatment. Occasionally, derepression of resistance genes may be the cause of the selection of resistant clones within the large bacterial inoculum associated with most pneumonias. Acinetobacter, Stenotrophomonas maltophilia, and Burkholderia cepacia are intrinsically resistant to many of the empirical antibiotic regimens listed in Table 251-8. VAP caused by these pathogens emerges during treatment of other infections, and resistance is always evident at initial diagnosis.
Table 251-8 Empirical Antibiotic Treatment of Health Care–Associated Pneumonia
Patients without Risk Factors for MDR Pathogens
Ceftriaxone (2 g IV q24h) or
Moxifloxacin (400 mg IV q24h), ciprofloxacin (400 mg IV q8h), or levofloxacin (750 mg IV q24h) or
Ampicillin/sulbactam (3 g IV q6h) or
Ertapenem (1 g IV q24h)
Patients with Risk Factors for MDR Pathogens
1. A -lactam:
Ceftazidime (2 g IV q8h) or cefepime (2 g IV q8–12h) or
Piperacillin/tazobactam (4.5 g IV q6h), imipenem (500 mg IV q6h or 1 g IV q8h), or meropenem (1 g IV q8h) plus
2. A second agent active against gram-negative bacterial pathogens:
Gentamicin or tobramycin (7 mg/kg IV q24h) or amikacin (20 mg/kg IV q24h) or
Ciprofloxacin (400 mg IV q8h) or levofloxacin (750 mg IV q24h) plus
3. An agent active against gram-positive bacterial pathogens:
Linezolid (600 mg IV q12h) or
Vancomycin (15 mg/kg, up to 1 g IV, q12h)
Note: MDR, multidrug-resistant.
Recommended options for empirical therapy are listed in Table 251-8. Treatment should be started once diagnostic specimens have been obtained. The major factor in the selection of agents is the presence of risk factors for MDR pathogens. Choices among the various options listed depend on local patterns of resistance and the patient's prior antibiotic exposure.
The majority of patients without risk factors for MDR infection can be treated with a single agent. The major difference from CAP is the markedly lower incidence of atypical pathogens in VAP; the exception is Legionella, which can be a nosocomial pathogen, especially when there are deficiencies in the treatment of a hospital's potable water supply.
The standard recommendation for patients with risk factors for MDR infection is for three antibiotics: two directed at P. aeruginosa and one at MRSA. The choice of a -lactam agent provides the greatest variability in coverage, yet the use of the broadest-spectrum agent—a carbapenem—still represents inappropriate initial therapy in 10–15% of cases.
Once an etiologic diagnosis is made, broad-spectrum empirical therapy can be modified to address the known pathogen specifically. For patients with MDR risk factors, antibiotic regimens can be reduced to a single agent in more than half of cases and to a two-drug combination in more than one-quarter. Only a minority of cases require a complete course with three drugs. A negative tracheal-aspirate culture or growth below the threshold for quantitative cultures, especially if the sample was obtained before any antibiotic change, strongly suggests that antibiotics should be discontinued. Identification of other confirmed or suspected sites of infection may require ongoing antibiotic therapy, but the spectrum of pathogens (and the corresponding antibiotic choices) may be different from those for VAP. If the CPIS decreases over the first 3 days, antibiotics should be stopped after 8 days. An 8-day course of therapy is just as effective as a 2-week course and is associated with less frequent emergence of antibiotic-resistant strains.
The major controversy regarding specific therapy for VAP concerns the need for ongoing combination treatment of Pseudomonas infection. No randomized controlled trials have demonstrated a benefit of combination therapy with a -lactam and an aminoglycoside, nor have subgroup analyses in other trials found a survival benefit with such a regimen. The unacceptably high rates of clinical failure and death for VAP caused by P. aeruginosa despite combination therapy (see "Failure to Improve," below) indicate that better regimens are needed—including, perhaps, aerosolized antibiotics.
VAP caused by MRSA is associated with a 40% clinical failure rate when treated with standard-dose vancomycin. One proposed solution is the use of high-dose individualized treatment, but the risk-to-benefit ratio of this approach is not known. Linezolid appears to be more efficacious than the standard dose of vancomycin, especially in patients with renal insufficiency.
Failure to Improve
Treatment failure is not uncommon in VAP, especially in that caused by MDR pathogens. In addition to the 40% failure rate for MRSA infection treated with vancomycin, VAP due to Pseudomonas has a 50% failure rate, no matter what the regimen. The causes of clinical failure vary with the pathogen(s) and the antibiotic(s). Inappropriate therapy can usually be minimized by use of the recommended triple-drug regimen (Table 251-8). However, the emergence of -lactam resistance during therapy is an important problem, especially in infection with Pseudomonas and Enterobacter spp. Recurrent VAP caused by the same pathogen is possible because the biofilm on ETs allows reintroduction of the microorganism. However, studies of VAP caused by Pseudomonas show that approximately half of recurrent cases are caused by a new strain. Inadequate local levels of vancomycin are the likely cause of treatment failure in VAP due to MRSA.
Treatment failure is very difficult to diagnose. Pneumonia due to a new superinfection, the presence of extrapulmonary infection, and drug toxicity must be considered in the differential diagnosis of treatment failure. Serial CPIS appears to track the clinical response accurately, while repeat quantitative cultures may clarify the microbiologic response. A persistently elevated or rising CPIS value by day 3 of therapy is likely to indicate failure. The most sensitive component of the CPIS is improvement in oxygenation.
Apart from death, the major complication of VAP is prolongation of mechanical ventilation, with corresponding increases in length of stay in the ICU and in the hospital. In most studies, an additional week of mechanical ventilation because of VAP is common. The additional expense of this complication warrants costly and aggressive efforts at prevention.
In rare cases, some types of necrotizing pneumonia (e.g., that due to P. aeruginosa) result in significant pulmonary hemorrhage. More commonly, necrotizing infections result in the long-term complications of bronchiectasis and parenchymal scarring leading to recurrent pneumonias. The long-term complications of pneumonia are underappreciated. Pneumonia results in a catabolic state in a patient already nutritionally at risk. The muscle loss and general debilitation from an episode of VAP often require prolonged rehabilitation and, in the elderly, commonly result in an inability to return to independent function and the need for nursing home placement.
Clinical improvement, if it occurs, is usually evident within 48–72 h of the initiation of antimicrobial treatment. Because findings on chest radiography often worsen initially during treatment, they are less helpful than clinical criteria as an indicator of clinical response in severe pneumonia. Although no hard and fast rules govern the frequency of follow-up chest radiography in seriously ill patients with pneumonia, assessment every few days in a responding patient seems appropriate. Once the patient has improved substantially and has stabilized, follow-up radiographs may not be necessary for a few weeks.
VAP is associated with significant mortality. Crude mortality rates of 50–70% have been reported, but the real issue is attributable mortality. Many patients with VAP have underlying diseases that would result in death even if VAP did not occur. Attributable mortality exceeded 25% in one matched-cohort study. Patients who develop VAP are at least twice as likely to die as those who do not. Some of the variability in reported figures is clearly related to the type of patient and ICU studied. VAP in trauma patients is not associated with attributable mortality, possibly because many of the patients were otherwise healthy before being injured. However, the causative pathogen also plays a major role. Generally, MDR pathogens are associated with significantly greater attributable mortality than non-MDR pathogens. Pneumonia caused by some pathogens (e.g., S. maltophilia) is simply a marker for a patient whose immune system is so compromised that death is almost inevitable.
(Table 251-6) Because of the significance of the ET as a risk factor for VAP, the most important preventive intervention is to avoid endotracheal intubation or at least to minimize its duration. Successful use of noninvasive ventilation via a nasal or full-face mask avoids many of the problems associated with ETs. Strategies that minimize the duration of ventilation have also been highly effective in preventing VAP.
Unfortunately, a tradeoff in risks is sometimes required. Aggressive attempts to extubate early may result in reintubation(s), which pose a risk of VAP. Heavy continuous sedation increases the risk, but self-extubation because of too little sedation is also a risk. The tradeoff is probably best illustrated by antibiotic therapy. Short-course antibiotic prophylaxis can decrease the risk of VAP in comatose patients requiring intubation, and data suggest that antibiotics decrease VAP rates in general. However, the major benefit appears to be a decrease in the incidence of early-onset VAP, which is usually caused by the less pathogenic non-MDR microorganisms. Conversely, prolonged courses of antibiotics consistently increase the risk of VAP caused by the more lethal MDR pathogens. Despite its virulence and associated mortality, VAP caused by Pseudomonas is rare among patients who have not recently received antibiotics.
Minimizing the amount of microaspiration around the ET cuff is also a strategy for avoidance of VAP. Simply elevating the head of the bed (at least 30° above horizontal but preferably 45°) decreases VAP rates. Specially modified ETs that allow removal of the secretions pooled above the cuff may also prevent VAP. The risk-to-benefit ratio of transporting the patient outside the ICU for diagnostic tests or procedures should be carefully considered, since VAP rates are increased among transported patients.
Emphasis on the avoidance of agents that raise gastric pH and on oropharyngeal decontamination has been diminished by the equivocal and conflicting results of more recent clinical trials. The role in the pathogenesis of VAP that is played by the overgrowth of bacterial components of the bowel flora in the stomach has also been downplayed. MRSA and the nonfermenters P. aeruginosa and Acinetobacter spp. are not normally part of the bowel flora but reside primarily in the nose and on the skin, respectively. Therefore, an emphasis on controlling overgrowth of the bowel flora may be relevant only in certain populations, such as liver transplant recipients and patients who have undergone other major intraabdominal procedures or who have bowel obstruction.
In outbreaks of VAP due to specific pathogens, the possibility of a breakdown in infection control measures (particularly contamination of reusable equipment) should be investigated. Even high rates of pathogens that are already common in a particular ICU may be a result of cross-infection. Education and reminders of the need for consistent infection control practices can minimize this risk.
While significantly less well studied than VAP, HAP in nonintubated patients, both inside and outside the ICU, is similar to VAP. The main differences are in the higher frequency of non-MDR pathogens and the better underlying host immunity in nonintubated patients. The lower frequency of MDR pathogens allows monotherapy in a larger proportion of cases of HAP than of VAP.
The only pathogens that may be more common in the non-VAP population are anaerobes. The greater risk of macroaspiration by nonintubated patients and the lower oxygen tensions in the lower respiratory tract of these patients increase the likelihood of a role for anaerobes. As in the management of CAP, specific therapy targeting anaerobes probably is not indicated unless gross aspiration is a concern.
Diagnosis is even more difficult for HAP in the nonintubated patient than for VAP. Lower respiratory tract samples appropriate for culture are considerably more difficult to obtain from nonintubated patients. Many of the underlying diseases that predispose a patient to HAP are also associated with an inability to cough adequately. Since blood cultures are infrequently positive (<15% of cases), the majority of patients with HAP do not have culture data on which antibiotic modifications can be based. Therefore, de-escalation of therapy is less likely in patients with risk factors for MDR pathogens. Despite these difficulties, the better host defenses in non-ICU patients result in lower mortality rates than are documented for VAP. In addition, the risk of antibiotic failure is lower in HAP.
American Thoracic Society/Infectious Diseases Society of America: Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med 171:388, 2005
Chastre J, Fagon JY: Ventilator-associated pneumonia. Am J Respir Crit Care Med 165:867, 2002 [PMID: 11934711]
Fagon JY et al: Invasive and noninvasive strategies for management of suspected ventilator-associated pneumonia. A randomized trial. Ann Intern Med 132:621, 2000 [PMID: 10766680]
Fine MJ et al: A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 336:243, 1997 [PMID: 8995086]
Lim WS et al: Defining community acquired pneumonia severity on presentation to hospital: An international derivation and validation study. Thorax 58:377, 2003 [PMID: 12728155]
Mandell LA et al: Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia. Clin Infect Dis 44(Suppl 2):S27, 2007
Singh N et al: Short-course empiric antibiotic therapy for patients with pulmonary infiltrates in the intensive care unit. A proposed solution for indiscriminate antibiotic prescription. Am J Respir Crit Care Med 162:505, 2000 [PMID: 10934078]
Vanderkooi OG et al: Predicting antimicrobial resistance in invasive pneumococcal infections. Clin Infect Dis 40:1288, 2005 [PMID: 15825031]