Pneumonia is inflammation of the lungs usually due to infection. There are two main types: lobar pneumonia and bronchopneumonia. Lobar pneumonia first affects one lobe of a lung. In the condition bronchopneumonia, inflammation ini-tially starts in the bronchi and bronchioles (airways), then spreads to affect patches of tissue in one or both lungs.

Causes  

Pneumonia is usually caused by various types of infection. Most cases are due either to viruses, such as adenovirus or respiratory syncytial virus, or to bacteria, such as Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus amd Mycoplasma pneumoniae. Another form is aspiration pneumonia, due to accidental inhalation of vomit. Aspiration pneumonia usually occurs in people whose cough reflex is not functioning, such as those who have drunk excessive amounts of alcohol or taken certain illegal drugs, or people who have suffered a head injury. Aspiration pneumonia can also occur in people with impaired swallowing who are elderly. This type of aspiration can occur silently. 

Symptoms and complications 

Symptoms usually include fever, chills, shortness of breath, a sharp chest pain, and a cough that produces yellow-green sputum and occasionally blood. Potential complications include pleural effusion (fluid around the lung), pleurisy (inflammation of the membrane lining the lungs and chest cavity), a lung abscess (collection of pus), and septicaemia (blood poisoning).

Diagnosis and treatment 

Diagnosis is made by physical examination, chest X-ray, and examining sputum and blood for microorganisms. Treatment depends on the cause, and usually includes antibiotic drugs. Aspirin or paracetamol may be given to reduce fever, and, in severe cases, oxygen therapy and artificial ventilation may be needed. In most cases, recovery usually occurs within two weeks.

Pneumonia in detail - technical

Essentials

Pneumonia is an acute or chronic infection involving the pulmonary parenchyma. Most cases are caused by microbial pathogens, usually bacteria or viruses and less often fungi or parasites. Pneumonia may also refer to inflammation involving the pulmonary parenchyma due to nonmicrobial causes such as chemical pneumonia. Other modifying terms are used as follows: pneumonia may be acute, subacute, or chronic, depending on the duration of symptoms; it may be described as bronchopneumonia, consolidated (lobar) pneumonia, interstitial pneumonia, or necrotizing pneumonia based on changes seen on chest radiography; or it may be named after the putative agent, e.g. pneumococcal pneumonia, mycoplasma pneumonia, pneumocystis pneumonia, etc. Pneumonia is also identified by the place of acquisition, e.g. community-acquired, nursing home-acquired, or hospital-acquired. This article is restricted to community-acquired pneumonia in the adult immunocompetent host.

Aetiology

Although the list of microbes that can cause pneumonia is legion, only a relatively small number are frequent pathogens, e.g. Streptococcus pneumoniae, Haemophilus influenzae, Mycoplasma pneumoniae, Chlamydia pneumoniae, legionella, anaerobic bacteria, and viruses. Less common pathogens are Moraxella catarrhalis, S. pyogenes, acinetobacter, C. psittaci, Coxiella burnetii, Neisseria meningitidis, Staphylococcus aureus, and enteric Gram-negative rods. In most reported series, each of the latter group generally accounts for less than 1 to 2% of cases. The relative frequencies of different pathogens causing community-acquired pneumonia in two large studies are summarized in Table 1. However, important limitations of these studies should be acknowledged: all the cases in the review conducted by the British Thoracic Society were inpatients, as were the great majority of those reviewed in the meta-analysis. Most studies of pneumonia show that only 20 to 30% of patients are sufficiently sick to require hospitalization. Furthermore, nearly all studies, including those that use extensive diagnostic resources, only identify a likely aetiological agent in 40 to 60% of cases. This suggests that fastidious microbes are under-represented, that many cases of pneumonia may be caused by as yet unidentified organisms, and that current diagnostic testing is pretty poor.

Epidemiology

Pneumonia is the most important infectious disease in terms of morbidity and mortality. It is estimated that in the United States of America there are 4 million cases of pneumonia per year (45 000 deaths), and worldwide there are 4400 million cases per year (4 million deaths). In the United States of America, data suggest that between 20 and 30% of all patients with a diagnosis of pneumonia are hospitalized, and that the mortality rate for this subpopulation is about 14%. The crude death rate from influenza and pneumonia in the United States of America for 1994 was 31.8 deaths per 100 000 of the population; this represents a 59% increase over the 20 deaths per 100 000 recorded in 1979, suggesting that the frequency of lethal pneumonia in the United States of America is increasing. Those aged 65 or older accounted for 89% of the deaths in 1994, suggesting that increases in longevity account for most of this increase in mortality rate.

Those pathogens associated with specific epidemiological and underlying conditions are summarized in Table 2. When an aetiological agent is identified, just three microbial agents account for most lethal cases of community-acquired pneumonia. Influenza accounts for an average of 20 000 deaths per year in the United States of America: the majority involve influenza A and occur in patients over 65 years of age, and most deaths are due to complications of influenza rather than influenza per se. The second common cause of lethal pneumonia is pneumococcal pneumonia; risk factors for a fatal outcome include bacteraemia, advanced age, and concurrent alcoholism. Legionella is the third agent, with associated mortality rates generally reported between 5 and 15% for patients with community-acquired infections.

Nearly all studies show that the risk of death with pneumonia is strongly associated with age extremes. Concurrent conditions that contribute to increased mortality rates include neoplastic disease, hepatic failure, congestive heart failure, cerebrovascular disease, and renal disease.

Pathogenesis

As with nearly all infectious diseases, the probability of disease depends on the virulence of the organism, the inoculum size, and the status of host defences. The normal tracheobronchial tree and lung parenchyma are sterile below the level of the larynx, so that agents of pneumonia must reach this site from external or adjacent sources, usually either by aspiration or inhalation. Organisms may also reach the lung by haematogenous seeding, direct extension from infection in a contiguous structure, or by activation of dormant organisms in the lung. These mechanisms are pathogen-specific, as summarized in Table 3.

Most bacterial pneumonias are probably caused by aspiration, which is defined as the abnormal entry of endogenous secretions or exogenous substances into the lower airways. There is a problem here with semantics because most cases of pneumonia are probably due to aspiration as classically described, but ‘aspiration pneumonia’ probably accounts for only 5 to 10% of cases. The explanation is presumably quantitative—‘aspiration’ generally referring to the abnormal entry of relatively large volumes in patients who are so predisposed due to dysphagia or a compromised level of consciousness. The alternative form is presumed to be ‘microaspiration’, involving the aspiration of very small numbers of microbes, a process that commonly takes place in healthy patients during sleep and with no apparent sequelae.

Clinical features

The classic presentation of pneumonia is of a cough and fever with the variable presence of sputum production, dyspnoea, and pleurisy. Most patients have constitutional symptoms such as malaise, fatigue, and asthenia, and many also have gastrointestinal symptoms. Although patients with pneumonia usually possess these characteristic clinical features, there can be major differences in presentation based on the host and the aetiological agent, as summarized below.

Pneumococcal pneumonia

S. pneumoniae is nearly always the most commonly identified pathogen in patients hospitalized with community-acquired pneumonia. A meta-analysis of 122 reports of community-acquired pneumonia for the period 1966 to 1995 showed that S. pneumoniae accounted for 65% of all cases where a microbial pathogen was defined and 66% of all bacteraemic cases. Studies using transtracheal aspiration or transthoracic aspiration—methods that avoid the problem of expectorated sputum contamination—show the presence of S. pneumoniae in 50 to 80% of cases.

The classic presentation is of a previously healthy adult with an upper respiratory tract infection who then develops a rigor followed by fever, dyspnoea, pleurisy, and a cough that usually becomes productive of a purulent, blood-streaked, or ‘rusty’ sputum. However, most patients show variations in this pattern, including one of a more subtle onset. Moreover, atypical presentations are particularly common in elderly patients.

Clinical examination can reveal features indicative of the severity of respiratory compromise—appearance of exhaustion, use of accessory muscles, inability to talk in sentences, tachypnoea (or even more worryingly when associated with exhaustion, a low respiratory rate), cyanosis—and of consolidation, in particular localized dullness to percussion and bronchial breathing.

Chest radiography nearly always shows an infiltrate, and lobar consolidation specifically suggests this diagnosis. A pleural effusion is present in about 25% of patients, but only 1 to 2% have an empyema. CT often shows lesions that are not apparent on chest radiographs.

Important observations over the past decade include the declining frequency of cases where this organism is identified, and the increasing resistance of S. pneumoniae to penicillin and a variety of other antibiotics. The declining frequency is commonly ascribed to a general decline in the quality of microbiological testing currently performed in cases of pneumonia in general, with greater dependence on rapid initiation of empirically selected antibiotics that are usually active against S. pneumoniae. Decreased antibiotic susceptibility is thought to reflect antibiotic abuse.

Poor prognostic findings in patients with pneumococcal pneumonia include advanced age, bacteraemia, alcoholism, and multiple lobe involvement.

The preferred antibiotics are amoxicillin for oral treatment and ceftriaxone or cefotaxime for parenteral treatment; penicillin-resistant strains may be treated with fluoroquinolones, vancomycin, or linezolid.

Haemophilus influenzae

This organism was originally described in 1892 by Pfeiffer who erroneously thought it was the agent of influenza; it was sometimes referred to as ‘Pfeiffer’s bacillus’. H. influenzae is always the second most common bacterial agent (behind S. pneumoniae) when an identified bacterial pathogen is found in community-acquired pneumonia. However, the diagnosis is difficult owing to problems with its recognition by direct Gram stain, the fastidious growth requirements of the organism, and interpretation—even when it is recovered—because it commonly colonizes the upper airways, leading to contamination of expectorated specimens. Type B H. influenzae is a well-established pathogen primarily in infants and young children, but is now a relatively rare cause of disease due to widespread use of H. influenzae (Hib) vaccine. H. influenzae strains causing pneumonia in adults are usually nontypable.

The clinical features are rather nonspecific and include fever, cough, purulent sputum, leucocytosis, and radiographic evidence of pneumonia—usually in a bronchopneumonic pattern, but it may occasionally be lobar. Patients with chronic obstructive lung disease often harbour H. influenzae in their lower airways and, allegedly, are prone to pneumonia caused by this organism, although supporting data for the association are not strong. Bacteraemia with H. influenzae in adults is infrequent. Most patients simply have a nonspecific pneumonia, with H. influenzae as the only potential pathogen identified in expectorated sputum.

About 30 to 45% of strains produce β-lactamase so that penicillin and amoxicillin are often ineffective. When H. influenzae is suspected or established the preferred agents are second- and third-generation cephalosporins, any combination of a β-lactam–β-lactamase inhibitor, azithromycin, or a fluoroquinolone.

Anaerobic bacteria

These organisms are the dominant components of the microbial flora in the upper airways and average 10¹²/ml in the gingival crevice. Anaerobes are the major pathogens identified in aspiration pneumonia and its sequelae, lung abscess and empyema. The major pathogens in this group are peptostreptcocci, bacteroides (other than Bacteroides fragilis), prevotella, and Fusobacterium nucleatum.

The typical patient has gingival crevice disease combined with a predisposition for aspiration that is usually due to a suppressed level of consciousness or dysphasia. The clinical presentation is usually more subtle than that for pneumococcal pneumonia in that the infection evolves over a period of many days, weeks, or even months. Chest radiographs usually show infection in a dependent segment (usually the superior segments of the lower lobes or posterior segments of the upper lobes since these are dependent in the recumbent position), fever, sputum that is often putrid, and evidence of chronic disease with weight loss or anaemia. Putrid discharge is very characteristic and diagnostic of anaerobic bacterial infection.

Aspiration pneumonia may also be due to chemical insults from gastric acid or other toxins, or may reflect the aspiration of foreign bodies or fluids (as in victims of drowning). However, the most common sequel to aspiration is bacterial infection involving the anaerobes that normally colonize the upper airways, and such bacteria account for 60 to 80% of cases of aspiration pneumonia, lung abscess, and, in many case series, empyemas. Although the bacterial aetiology can be identified from anaerobic cultures of uncontaminated specimens, these are generally not obtained except in the case of pleural fluid in the presence of empyema; even then the cultures are often falsely negative due to inadequate techniques used to recover oxygen-sensitive bacteria. Thus, the aetiological diagnosis is usually based on the clinical features—where key clues are the chronicity of the infection, associated conditions suggesting aspiration, tissue necrosis with abscess formation, or a bronchopleural fistula leading to empyema and/or putrid discharge.

The preferred drugs are clindamycin or a β-lactam–β-lactamase inhibitor.

Mycoplasma pneumoniae

This organism is one of the most common causes of lower airways infection in young adults, and it is now more frequently recognized in older adults. The original appellation was ‘primary atypical pneumonia’, a term applied in the 1930s to a relatively benign form of pneumonia to distinguish it from pneumococcal pneumonia. Early work showed that it was associated with a serum factor that agglutinated erythrocytes in the cold; furthermore, Eaton reported that the infection was transmissible from person to person by intracheal inoculations. Thus, atypical pneumonia, cold-agglutinin pneumonia, and Eaton-agent pneumonia were found to be synonymous.

The typical patient is usually a young adult who experiences a respiratory tract infection accompanied by headache, myalgia, cough, and fever, and with a chest radiograph that shows bronchopneumonia. The cough is often nonproductive, but when sputum is obtained it is mucoid, shows predominantly mononuclear cells, and no dominant organism. A characteristic feature is the relatively high frequency of extrapulmonary complications such as rash, neurological syndromes (aseptic meningitis, encephalitis, neuropathies), myocarditis, pericarditis, and haemolytic anaemia. The diagnosis should be suspected in those patients with a relatively mild form of pneumonia, particularly in previously healthy young adults.

Most laboratories do not cultivate mycoplasma due to the effort needed to recover the organism, the long time required, and the ease of empirical treatment. Serological tests may be used, but their merits are disputed. Polymerase chain reaction (PCR) and other rapid diagnostic tests are under development.

With regard to treatment, the pathogen lacks a cell wall and hence is not susceptible to penicillin, cephalosporins, or other cell-wall active antibiotics. The usual therapeutic agents are macrolides (such as erythromycin, clarithromycin, or azithromycin) or doxycycline; fluoroquinolones are also active.

Chlamydia pneumoniae

This relatively recently identified pathogen is now thought to account for about 5 to 10% of all community-acquired cases of pneumonia, often in young adults who present in a fashion quite similar to that of patients with a mycoplasma pneumonia. C. pneumoniae continues to be regarded as a relatively benign agent of pneumonia: most patients have an upper airways infection with this organism, laryngitis is relatively common, bronchitis is less common, and pneumonitis is an infrequent complication. C. pneumoniae plays a role in exacerbations of asthma, and the organism was once thought to be involved in some chronic conditions such as cardiovascular disease, but more recent large controlled trials have not supported an association.

The diagnosis of chlamydia pneumonia is difficult. The organism is cultivated like a virus using tissue cultures, but few laboratories offer this test. Serology is difficult to interpret; the usual titres for IgM or serial changes with acute and convalescent sera are arbitrary. Like mycoplasma, this is an organism that is often suspected, infrequently proven, and easily treated empirically. A PCR test is expected soon.

The usual treatment is doxycycline, a macrolide (erythromycin, clarithromycin, or azithromycin), or a fluoroquinolone.

Legionella

Legionnaires’ disease was originally described during the American Legion Convention in Philadelphia in 1976, with the putative agent reported the following year. Legionella cause two major syndromes: the pneumonic form or legionnaires’ disease, referring to the American Legion Convention epidemic, and a benign influenza-like illness called ‘Pontiac fever’ in reference to an outbreak in 1967 in Pontiac, Michigan. Although legionnaires’ disease is often grouped with mycoplasma and chlamydia infection as being an ‘atypical pneumonia’, it is a quite different pulmonary infection because it occurs primarily in older adults, is a serious and often lethal form of pneumonia, and most hospital laboratories have diagnostic resources to establish the aetiology. Legionnaires’ disease is defined as pneumonia caused by any species of the genera legionella, but the great majority of cases are caused either by Legionella pneumophila (80 to 90% of cases) or L. mcdadei (5 to 10%). This disease may be epidemic or sporadic. Epidemics usually occur in buildings, especially hotels and hospitals, and they reflect legionella contamination of the potable water or cooling systems of air conditioners. Predisposing factors include exposure to environmental sources of legionella (there is no patient-to-patient transmission), age over 40 years, smoking, or reduced cell-mediated immune responses as with organ transplantation, cancer chemotherapy, or chronic corticosteroid usage; patients with AIDS do not seem to be uniquely susceptible.

There are no remarkable features of the clinical presentation, except that patients are almost invariably quite sick and may be critically ill. In addition to the typical symptoms of pneumonia with cough and dyspnoea, most present with a profound systemic illness with high fever and myalgias, often with gastrointestinal and neurological symptoms.

The diagnosis can be established with a urinary antigen assay for the detection of L. pneumophila serogroup I (which accounts for about 80% of cases), culture of respiratory secretions on selective media, or serology. All these tests are quite specific, but none are sufficiently sensitive to exclude the diagnosis when they are negative, and the urinary antigen assay is the only one that is easily done and gives results in a timely fashion.

The drugs of choice are fluoroquinolone, or a macrolide, or azithromycin, but mortality rate is generally reported to be 5 to 15% even with proper therapy.

Read about Legionnaires' disease in more detail

Staphylococcus aureus

Staphylococcal pneumonia was classically described as a complication of influenza during the 1918 epidemic of ‘Spanish flu’. This organism continues to be a potentially important superinfecting pathogen in influenza, and is the most common form of embolic pulmonary infection with injection drug use and tricuspid valve endocarditis. A relatively new form of Staphyloccocus aureus pneumonia involves strains with the Panton Valentine leukocidin (PVL) gene. These cases are most common in children or young adults with influenza, the clinical course is fulminant, often with pulmonary necrosis and leucopenia, and the mortality rate is high.

The organism can usually be recovered in blood cultures and in respiratory secretions. However, care must be exercised when interpreting respiratory secretion cultures that yield Staphyloccocus aureus since this may be a contaminant, and it is particularly common as a contaminant in those patients who have received previous antibiotic treatment.

The treatment should be based on in vitro susceptibility tests, usually an antistaphylococcal penicillin (flucloxacillin, oxacillin, or nafcillin), a first-generation cephalosporin (cefazolin), or vancomycin (for methicillin-resistant strains and for patients with severe penicillin allergy). The PVL-positive strains are usually methicillin-resistant and should be treated with vancomycin or linezolid.

Gram-negative bacilli

Klebsiella pneumoniae was originally described in 1882 by Friedlander, who believed it was the cause of pneumococcal pneumonia. This organism has continued to be a rare but important cause of community-acquired pneumonia, accounting for about 0.5 to 1.5% of all cases. The classic presentation of ‘Friedlander’s pneumonia’ was a serious pneumonia in an alcoholic patient with a chest radiograph that showed upper lobe involvement and the ‘bulging fissure sign’ (indicating abscess formation) and sputum that resembled currant jelly. This form of klebsiella pulmonary infection is rarely encountered now, although klebsiella infection is occasionally implicated in community-acquired pneumonia.

Other Gram-negative bacilli may also cause pneumonia, but the frequency in immunocompetent hosts is very low. Pseudomonas aeruginosa is a rare pulmonary pathogen, but should be suspected when recovered in respiratory secretions from patients with specific predisposing conditions including structural lung disease, neutropenia, cystic fibrosis, or advanced AIDS. Gram-negative bacteria are commonly encountered in cultures of respiratory secretions in patients who have already started antibiotic treatment, when care must be exercised in interpretation because they are usually contaminants reflecting upper airway colonization.

Treatment should be based on in vitro sensitivity tests.

Viruses

Viral infections of the lower airways account for pneumonia in 10 to 15% of inpatients, and probably a substantially larger number of those managed as outpatients. The most frequent pathogens are influenza, parainfluenza, and respiratory syncytial virus (RSV). Influenza infections with bronchitis occur in epidemics, but influenza pneumonia is rare. More common in influenza patients with chest radiographs showing infiltrates is bacterial superinfection, most frequently with S. pneumoniae or Staphylococcus aureus; less common superinfecting pathogens in this setting are N. meningiditis and group A streptococcus. The diagnosis of influenza can be made by the combination of an established epidemic and typical influenza symptoms, especially fever. The alternative is to establish the presence of the organism by one of several rapid tests for influenza-A or influenza-B antigen that provide results that are available in about 20 min, have a sensitivity of about 70 to 80%, and excellent specificity (in epidemics).

Clinical features of influenza are generally well known and include cough, fever, purulent sputum, and myalgias. Patients with bacterial superinfections will usually have typical influenza-like symptoms, improve, and then deteriorate after 1 to 2 weeks.

Infections involving influenza A or B may be treated with the neuraminidase inhibitors zanamivir or oseltamivir. If given within 48 h of the onset of symptoms these reduce the duration of typical symptoms by 1 to 1.5 days. They are more effective in seriously ill patients and when given very early in the 48-h window. Amandidine and ramantidine are no longer commonly advocated for influenza due to high rates of resistance.

RSV has usually been considered a pathogen in children but is now recognized with increasing frequency in adults, especially older people. The diagnosis is easily established with a direct fluorescent mononoclonal antibody (DFA) stain of respiratory secretions in children, but this test is much less sensitive in adults. Ribavirin is active against RSV and is sometimes used by inhalation therapy in children, but the benefit of this treatment is debated.

Bullet list 1 Recommended laboratory tests in suspected community-acquired pneumonia

Laboratory diagnosis

Laboratory tests are used to establish the diagnosis, evaluate the severity, and identify the aetiological agent (Bullet list 1above).

Tests to establish the diagnosis and evaluate severity

Chest radiography

The chest radiograph is a pivotal test for the confirmation of pneumonia, it being impossible to make this diagnosis in the absence of a new infiltrate, with four possible exceptions: (1) dehydration, (2) neutropenia, (3) early in the course, or (4) pneumocystis pneumonia (PCP). None is common or verified, excepting PCP, which may show no infiltrate in up to 30% of cases. In all cases CT is more sensitive in detecting infiltrates and in defining pathology, and is indicated in unusual cases.

Most patients with symptoms of pneumonia and a negative chest radiograph have acute bronchitis, which is generally caused by viral pathogens that do not respond to antibiotic treatment. Thus, the importance of the chest radiograph is in confirming pneumonia, which is a critical feature in avoiding antibiotic abuse. Additional advantages of the chest radiograph are that it provides assistance in identifying the aetiological agent, establishes a baseline for subsequent evaluation, provides prognostic information, and permits the detection of underlying or associated conditions such as a neoplasm.

Other laboratory tests

The most useful additional laboratory tests to determine the severity of illness and need for hospitalization are evaluation of blood oxygenation with pulse oximetry or arterial blood gas determination, blood chemistries (glucose, blood urea nitrogen, and serum sodium levels), and a full blood count. Depending on context, patients who are hospitalized for pneumonia should generally undergo serological testing for HIV (after appropriate consent), since this is a common predisposing cause.

Studies to determine microbial aetiology

Laboratory studies for identifying pulmonary pathogens are among the most controversial issues in pneumonia management. Most physicians now have a nihilistic approach, concluding that microbial studies in cases of pneumonia are usually negative, are not cost-effective, and are largely unnecessary since empirical treatment is generally successful. The guidelines from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS) emphasize the best indications for microbiological studies are for patients sufficiently sick to require hospitalization in an intensive care unit, cases where the cause is a pathogen not covered by standard empirically selected treatment, and for some specific agents that have important implications for prognosis and epidemiology such as legionellosis, Staphylococcus aureus, pandemic influenza, agents of bioterrorism, and Gram-negative bacilli (although several of these circumstances cannot be determined in advance of microbiological testing). However, it should be emphasized that Gram stain and culture of expectorated sputum is never ‘wrong’, especially if done with good quality control.

Although empirical therapy is generally advocated for outpatients, routine microbiological testing to identify the aetiological agent of pneumonia is generally recommended for inpatients. Such tests include blood cultures (from blood samples taken prior to the initiation of antibiotic treatment), which yield a pathogen in about 12% of cases. In general, the only additional test commonly performed to identify an aetiological agent is an expectorated sputum Gram stain and culture. Practice standards for this process include the following:

• ◆ The specimen should be obtained by deep cough and should be grossly purulent. It should be collected before antibiotic therapy, preferably in the presence of a physician or nurse.
• ◆ The specimen should be promptly transported to the laboratory for processing and incubation within 2 to 5 h.
• ◆ A qualified technician should select a purulent portion for Gram stain and culture.
• ◆ Cytological screening should be done under low-power magnification (×100) to determine cellular composition as a contingency for culture.
• ◆ The sample should be cultured using standard techniques, with results reported by semiquantitative assessment; most pathogens are recovered in moderate or heavy growth, indicating more than five colonies in the second streak.
• ◆ Interpretation should be based on the correlation of the Gram stain, semiquantitative culture results, and clinical observations.

The aetiological agent of pneumonia is considered to be clearly established if a likely pulmonary pathogen is recovered from an uncontaminated specimen such as blood culture, pleural fluid, transtracheal aspiration, or transthoracic aspirate. Alternatively, the very presence of a likely pathogen recovered from respiratory secretions is tantamount to a diagnosis; organisms in this category include legionella species, Mycobacterium tuberculosis, most viruses other than the herpesvirus group (influenza virus, respiratory syncytial virus, Hantavirus, parainfluenza virus, and adenovirus), and certain fungi (Histoplasma capsulatum, Coccidioides immitis, Blastomyces dermatitidis, and Pneumocystis jirovecii). Organisms such as S. pneumoniae, M. catarrhalis, H. influenzae, and Staphylococcus aureus may be pulmonary pathogens, but interpretation is problematic due to possible contamination with specimens from the upper airway flora. Organisms that virtually never represent pulmonary pathogens include S. epidermidis, enterococcus, neisseria other than N. meningitidis, candida, and Gram-positive bacilli other than nocardia or actinomyces.

Transtracheal aspiration was once a popular method of obtaining specimens from the lower airways that avoided upper airway contamination, but the technique requires a skilled clinician and is generally thought to be too invasive for routine use. Transthoracic aspiration has the same limitations, and furthermore seems to give a relatively large number of false-negative results. Bronchoscopy is an attractive method for obtaining respiratory secretions directly from the lower airways; however, the procedure is complicated by contamination with instrument passage through the upper airways so that routine cultures of bronchoscopic aspirates are no better than expectorated sputum. These results may be substantially improved with quantitative cultures of bronchoalveolar lavage specimens or quantitative brush specimens, but many laboratories do not offer this type of analysis, and many pulmonary services cannot provide the samples in a timely fashion.

Most hospital laboratories offer diagnostic tests for detecting legionella spp. using urinary antigen assay and culture. Urinary antigen testing is advocated because it is rapid, simply performed, and highly specific; disadvantages include the fact that it only detects L. pneumophila serogroup 1, although this accounts for 80% of cases. The alternative test for detecting legionella is culture, which has the advantage of detecting all species of legionella, but the disadvantage that it requires 3 days, requires specialized media, and is technically demanding. A new urinary antigen assay diagnostic test for S. pneumoniae can provide results within 3 h.

Most laboratories do not offer diagnostic tests to detect M. pneumoniae or C. pneumoniae, despite their presumed frequency. This reflects the lack of an acceptable test that is easily performed, provides adequate sensitivity and specificity, and can provide results in a timely fashion.

Treatment

Critical components of initial treatment may include intravenous hydration, oxygenation, and/or intubation and mechanical ventilatory support. Pleural effusions should be sampled to exclude empyema and, when the effusions are large, drained to improve oxygenation. Most authorities feel that expectorants, cough suppressants, and chest physiotherapy are of little value.

Antibiotic therapy

Antibiotics are the mainstay of therapy. Suggestions for specific agents according to microbial pathogen are summarized in Table 4. Most of these are relatively noncontroversial and demonstrate the advantage of establishing an aetiological agent. However, as noted above, no pathogen can be detected in 40 to 60% of cases despite arduous attempts to do so, and even when an agent is found, this information is usually not available when initial therapeutic decisions are needed. For this reason, most patients are treated empirically, at least initially, whilst microbiological results are pending. Recommendations for empirical treatment are summarized in Table 5. These options are selected on the basis of predicted activity against the most likely pathogens and extensive clinical trials. Nevertheless, this is one of the most controversial areas in medicine based on concerns about antibiotic abuse, increasing resistance of S. pneumoniae to many antimicrobials, and sharp geographical differences in the rates of S. pneumoniae resistance.

Timing of antibiotic therapy

A retrospective trial of over 20 000 Medicare patients in the United States of America hospitalized with community-acquired pneumonia showed that mortality increased with a progressive delay in the time taken to initiate antibiotic therapy after patients had been evaluated. The increase in mortality became statistically significant when the delay exceeded 6 h. This observation is not surprising since pneumonia is a potentially lethal infection that usually responds to antibiotics, so any delay in treatment would be expected to have deleterious effects. As a result of these observations, many hospitals in the United States of America are now audited to determine their compliance with antibiotic recommendations according to ATS/IDSA guidelines and initiation of this treatment within 8 h of a patient’s admission to the Emergency Department or hospital.

Monitoring response to therapy

Subjective responses are usually noted within 3 to 5 days of initiating treatment. Objective parameters to monitor include fever, oxygen saturation, peripheral leucocyte count, and changes on serial chest radiographs. The most carefully documented responses are mortality rates, time to defervescence, and duration of hospital stay. With regard to fever, the temperature in young adults with pneumococcal pneumonia usually drops within 2 to 3 days, whereas those with bacteraemic pneumococcal pneumonia (usually elderly patients) respond more slowly. Blood cultures in bacteraemic patients are usually negative within 24 to 48 h. Cultures of sputum will usually show eradication of bacterial pathogens within 24 to 48 h, a major exception being P. aeruginosa. Radiographic appearances are slow to improve and much less useful than clinical observations for evaluating response. Follow-up radiographs are generally not recommended, except for patients who are over 40 years of age or are smokers, and the suggested time to do this is 7 to 12 weeks after initiating treatment.

Patients who are initially treated with intravenous antibiotics can usually be changed to receive oral agents when they are able to take oral medications and show clinical improvement, such as a temperature below 38°C for 24 h, a respiratory rate of less than 24/min, and when the P O ₂ has returned to normal.

Failure to respond

The major considerations in patients who fail to respond according to the guidelines noted above are:

• ◆ The disease is too far advanced at the time of treatment, or treatment is delayed for too long: this is most commonly seen with pneumonia caused by S. pneumoniae or legionella.
• ◆ The wrong antibiotic was selected: this is uncommon, but in our experience the most common exceptions are tuberculosis, pneumocystis pneumonia, and viral pneumonias.
• ◆ An inadequate antibiotic dosage is given.
• ◆ The wrong diagnosis is made: for example, there is a noninfectious disease such as pulmonary embolism with infarction, congestive failure, Wegener’s granulomatosis, sarcoidosis, atelectasis, chemical pneumonitis, or bronchiolitis obliterans organizing pneumonia.
• ◆ The wrong microbial diagnosis is made.
• ◆ The patient may be debilitated, have a severe associated disease, or be immunosuppressed, or there may be other host inadequacies.
• ◆ There may be a complicated pneumonia with undrained empyema, metastatic site of infection (meningitis), or bronchial obstruction (foreign body, carcinoma).
• ◆ There may be a pulmonary superinfection: most patients in this category respond and then deteriorate with a new fever.
• ◆ There may be a complication of the antibiotic treatment such as an adverse drug reaction or antibiotic-associated colitis.

Prognosis

The overall mortality for patients who are hospitalized with community-acquired pneumonia, according to a meta-analysis of 122 reports, is 14%. Risk factors for lethal outcome were well described in the prepenicillin era, when extremes of age were probably the most important factor. Other risks included bacteraemia, the extent of changes on chest radiography, alcohol consumption, and the extent of leucocytosis. More recent studies have continued to show that these factors, especially age, are major risk factors for morbidity and mortality. Investigators from the Pneumonia Patient Outcomes Research Team (PORT) have developed a prediction rule using a cumulative point score obtained from five categories comprising 19 variables (Table 18.4.2.6). This prediction rule was applied retrospectively to 38 039 inpatients and showed a direct correlation between numerical score and mortality, the authors concluding that these factors predict outcome and can also be used to determine the need for hospitalization or the need for the intensive care unit.

An alternative metric that is simpler to remember and equally good for prediction is the six-point ‘CURB-65’ score, based on Confusion (impaired consciousness), elevated blood Urea nitrogen (>7 mmol/litre), increased Respiratory rate (>30/min), low systolic Blood pressure (<90 mmHg), low diastolic Blood pressure (<60 mmHg), and age over 65 years. In over 1000 prospectively studied patients with community-acquired pneumonia from three countries (United Kingdom, New Zealand, the Netherlands), the risk of mortality or need for intensive care admission was as follows: score 0, 0.7%; score 1, 3.2%; score 2, 13%; score 3, 17%; score 4, 41.5%; score 5, 57%. If the blood urea nitrogen level is not available and only the clinical parameters are considered (CRB-65), the risk of mortality for particular scores (out of a maximum of 5) was as follows: score 0, 1.2%; score 1, 5.3%; score 2, 12.2%; score 3, 32.9%; score 4, 18.2%. These data have led to recommendations from the British Thoracic Society that patients who have a CRB-65 score of 0 do not normally require hospitalization for clinical reasons; referral to and assessment in hospital should be considered for those with a score of 1 or (particularly) 2; and those with a score of 3 or more require urgent hospital admission (unless known to be terminally ill).

With regard to specific pathogens, the main agents of community-acquired pneumonia associated with high mortality rates are bacteraemic pneumococcal pneumonia and legionnaires’ disease. Influenza is directly or indirectly implicated in 20 000 to 40 000 deaths per year in the United States of America, but primary influenza pneumonia is relatively rare and most of the influenza-associated deaths are of elderly patients who succumb to complications of influenza. It should also be noted that pneumonia is an extremely common terminal event in patients who die of other conditions, presumably because of aspiration in the terminal stages. Thus, pneumonia is a common autopsy finding when other medical conditions are actually the primary cause of death.

Prevention and control

The main preventive measures are influenza and S. pneumoniae vaccination. The components selected for the influenza vaccine each year are based on the anticipated strains for the forthcoming season, a prediction that has been quite accurate in 14 of the 16 influenza seasons from 1989/90 through 2006/07. Protective efficacy is generally 60 to 70% in the general population when there is a good match between the vaccine strains and the epidemic strain; it is less effective in elderly vaccinees, but those who develop influenza after vaccination usually have an attenuated course with significant reduction in mortality. The current recommendation is for vaccination between October and November of patients living in the northern hemisphere. Targeted populations are summarized in Bullet list 2. Zanamivir and oseltamivir may be used to prevent influenza in unvaccinated patients who are so exposed. Amantidine and rimantidine are no longer recommended due to resistance.

The 23-valent vaccine for S. pneumoniae contains capsular polysaccharide from 23 serogroups that are responsible for 80 to 85% of bacteraemic pneumococcal infections. Studies of this vaccine suggest a 60% efficacy in preventing bacteraemic pneumococcal infection, but efficacy in reducing rates of community-acquired pneumonia or even pneumococcal pneumonia is not consistently shown. A more recently developed 7-valent protein-conjugated pneumococcal vaccine has the advantage of stimulating a good antibody response in children under 2 years of age. This has not been extensively tested in adults, but its use in children has resulted in a 50 to 80% decrease in rates of invasive pneumococcal infections in adults, the implication being that children less than 2 years are main vectors of pneumococcal infections (as they probably are for influenza).

Bullet list 2 Recommendation for influenza vaccination

Controversies

There are probably few diseases in medicine that have been better studied than pneumonia, but with such extraordinary controversy in management guidelines, including the utility of microbiology studies, the empirical selection of antibiotics, and the use of pneumococcal vaccine. Emergency of "replacement serotypes such as 19A has resulted in a new vaccine, Prevnar 13.

Studies of microbial aetiology

Culture and Gram stain of expectorated sputum is the time-honoured method for determining the microbiology of community-acquired pneumonia. Nevertheless, there is substantial controversy regarding the worth of this exercise and a wealth of medical reports with highly divergent findings that simply fuel the debate. In general, the best results were achieved in the prepenicillin era, when sputum bacteriology was an art and many patients underwent transthoracic needle aspiration to be sure the pathogen was found, the reason being that the only available therapy was type-specific antisera for S. pneumoniae, hence treatment required retrieval of the specific strain. High-quality laboratory technology persisted through the mid-1980s, when the yield of S. pneumoniae in expectorated sputum samples for inpatients with community-acquired pneumonia was generally reported at 40 to 70%. More recent experience is much different, with the yield of S. pneumoniae in expectorated sputum by either Gram stain or culture being only 5 to 10% in most studies of large medical systems such as Medicare. Arguments favouring sputum microbiology are the benefits of pathogen-directed therapy that restrains antibiotic abuse, limits side effects, and reduces cost. In addition, this permits the identification of epidemiologically important organisms, knowledge of which provides the database for empirical therapy recommendations. Arguments against microbiological studies include the facts that this procedure, as currently performed in most laboratories, shows a low yield, the information is infrequently available when therapeutic decisions are made, empirical treatment is usually effective against the most common pathogens, and—even if a pathogen is recovered—there is no good way to exclude the presence of a copathogen. Many authorities now feel molecular diagnostics will supplant traditional microbiology methods.

Antibiotic selection

The selection of antimicrobials is usually easy if the pathogen is known, but more difficult with empirical decisions when it is not. There are many ‘trade-offs’ with empiricism, including the consequence of promoting resistance, side effects, and cost. There is also controversy about the pathogens that need to be covered empirically, a major issue being whether or not it is necessary to treat ‘atypical agents’.

Meta-analyses of proper trials show that β-lactams are as effective as regimens with activity against Mycoplasma pneumoniae and C. pneumoniae, although legionella is the exception. Thus, in many countries the standard for outpatient treatment is amoxicillin, and for the United States of America and some European countries it is doxycycline or a macrolide. Fluoroquinolones are highly effective against most treatable bacterial pathogens, but there is concern that excessive usage will lead to costs in the form of resistance and increased incidence of C. difficile. The ATS/IDSA 2007 guidelines recommend these agents only when penicillin-resistant S. pneumoniae is suspected, the major risks for which are antibiotic exposure within the previous 3 months, or patients with important comorbidities such as diabetes, chronic renal failure, or serious cardiopulmonary disease.

For hospitalized patients the ATS/IDSA recommendations are based on the Medicare database, which was analysed for mortality rates with different regimens for adults who are hospitalized for community-acquired pneumonia. These recommendations are (1) a fluoroquinolone (levofloxacin or moxifloxacin), or (2) a macrolide (usually azithromycin) combined with a cephalosporin (ceftriaxone or cefotaxime). These regimens reduced mortality in the analysis of over 14 000 Medicare patients by 36% and 24%, respectively.

Some of the controversies that have emerged include: the relative merits of β-lactams; the variable rates of β-lactam resistance by S. pneumoniae; the place of erythromycin in guidelines (based on price, tolerance, and activity vs legionella); the need to provide ‘double coverage’ (β-lactam plus a macrolide) in patients with pneumococcal bacteremia; the role of pathogen-directed therapy in the era of poor-quality microbiology; conclusiveness of the urinary antigen assay for S. pneumoniae; and the following important but often neglected exceptions to the recommendations:

• ◆ Influenza with suspected bacterial superinfection: in these cases the major pathogens are S. pneumoniae and Staphylococcus aureus, which need prioritization in antibiotic selection
• ◆ The immunocompromised patient, where the list of pathogens is legion and driven by multiple variables including characteristics of the immunological deficit
• ◆ Aspiration pneumonia, which may be bacterial, usually anaerobes and/or streptococci, or might be chemical or obstructive due to a foreign body
• ◆ ‘Health care associated’, with reference to people who are not in hospital but often have pneumonia that may be more similar to pathogens that cause nosocomial pneumonia than those encountered in community-acquired pneumonia. This category includes patients in chronic care facilities, dialysis centres, and other extensions of the health care system

Pneumococcal vaccine

The polysaccharide vaccine has established merit in reducing pneumococcal bacteraemia, but most prospective randomized controlled trials have failed to show a significant benefit in terms of reducing the rates of pneumonia or rates of pneumococcal pneumonia. Most reports of a beneficial effect of vaccination have been based on statistical analyses of the serotype of patients with pneumococcal infection, which demonstrated higher rates of vaccine strains in unvaccinated patients. Even these studies failed to show a benefit in the highest risk group, namely the elderly and the immunosuppressed. The need for a pneumococcal vaccine is widely appreciated due to the extent of morbidity and mortality caused by S. pneumoniae and the increasing difficulty caused by resistance in treating these infections. Many authorities feel that the best solution to the dilemma is a better pneumococcal vaccine. The best data for benefit at present is widespread use of the protein-conjugated vaccine, such as Prevnar 13, given to children before 2 years of age to protect themselves and adults.

Further reading