Table 32.1 gives possible aetiological agents based on epidemiological clues. Streptococcus pneumoniae is the most commonly isolated organism. The next most common pathogens in patients admitted to ICU are: Legionella species, Haemophilus influenzae, Enterobacteriaceae species, Staphylococcus aureus and Pseudomonas species.¹⁰

Table 32.1 Possible aetiological agents based on epidemiological clues ^2,^3,^7,⁹

Exposure	Organism
Exposure to animals
Handling turkeys, chickens, ducks or psittacine birds or their excreta	Chlamydia psittaci
Exposure to birds in countries in which avian flu has been identified in birds	Influenza A H5N1
Handling infected parturient cats, cattle, goats or sheep or their hides	Coxiella burnetii
Handling infected wool	Bacillus anthracis
Handling infected cattle, pigs, goats or sheep or their milk	Brucella spp.
Insect bite. Transmission from rodents and wild animals (e.g. rabbits) to laboratory workers, farmers and hunters	Francisella tularensis
Insect bites or scratches. Transmission from infected rodents or cats to laboratory workers and hunters	Yersinia pestis
Contact with infected horses (very rare)	Pseudomonas mallei
Exposure to mice or mice droppings	Hantavirus
Geographical factors
Immigration from or residence in countries with high prevalence of tuberculosis	Mycobacterium tuberculosis
North America. Contact with infected bats or birds or their excreta. Excavation in endemic areas	Histoplasma capsulatum
South-west USA	Coccidiodes species, Hantavirus
USA. Inhalation of spores from soil	Blastomyces dermatitidis
Asia, Pacific, Caribbean, north Australia. Contact with local animals or contaminated skin abrasions	Burkholderia pseudomallei
Host factors
Diabetic ketoacidosis	Streptococcus pneumoniae, Staphylococcus aureus
Alcoholism	Streptococcus pneumoniae, Staphylococcus aureus, Klebsiella pneumoniae, oral anaerobes, Mycobacterium tuberculosis, Acinetobacter spp.
Chronic obstructive pulmonary disease or smoking	Streptococcus pneumoniae, Haemophilus influenzae, Moraxella catarrhalis, Chlamydia pneumoniae, Legionella spp., Pseudomonas aeruginosa
Sickle-cell disease	Streptococcus pneumoniae
Pneumonia complicating whooping cough	Bordatella pertussis
Pneumonia complicating influenza	Streptococcus pneumoniae, Staphylococcus aureus
Pneumonia severe enough to necessitate artificial ventilation	Streptococcus pneumoniae, Legionella spp., Staphylococcus aureus, Haemophilus influenzae, Mycoplasma pneumoniae, enteric Gram-negative bacilli, Chlamydia pneumoniae, Mycobacterium tuberculosis, viral infection, endemic fungi
Nursing-home residency	Treat as health care-associated pneumonia
Poor dental hygiene	Anaerobes
Suspected large-volume aspiration	Oral anaerobes, Gram-negative enteric bacteria
Structural disease of lung (e.g. bronchiectasis, cystic fibrosis)	Pseudomonas aeruginosa, Burkholderia cepacia, Staphylococcus aureus
Lung abscess	Community-acquired methicillin-resistant Staphylococcus aureus, oral anaerobes, endemic fungi, Mycobacterium tuberculosis, atypical mycobacteria
Endobronchial obstruction	Anaerobes, Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus
Intravenous drug addict	Staphylococcus aureus, anaerobes, Mycobacterium tuberculosis, Streptococcus pneumoniae
Others
Epidemic	Mycoplasma pneumoniae, influenza virus
Air-conditioning cooling towers, hot tubs or hotel or cruise ship stay in previous 2 weeks	Legionella pneumophilia
Presentation of a cluster of cases over a very short period of time	Bioterrorist agents: Bacillus anthracis, Franciscella tularensis, Yersinia pestis

CLINICAL PRESENTATION

Pneumonia produces both systemic and respiratory manifestations. Common clinical findings include fever, sweats, rigors, cough, sputum production, pleuritic chest pain, dyspnoea, tachypnoea, pleural rub and inspiratory crackles. Classic signs of consolidation occur in fewer than 25% of cases. Multiorgan dysfunction or failure may occur depending on the type and severity of pneumonia.

The diagnosis of pneumonia may be more difficult in the elderly. Although the vast majority of elderly patients with pneumonia have respiratory symptoms and signs, over 50% may also have non-respiratory symptoms and over one-third may have no systemic signs of infection.

INVESTIGATIONS ^7,⁸

Investigations should not delay administration of antibiotics as delays are associated with an increase in mortality.³ Important investigations include:

1 CXR

2 Arterial blood gases or oximetry

3 Full blood count

4 Serum creatinine, urea and electrolytes

5 Liver function tests

6 Blood cultures (×2) prior to the administration of antimicrobials

7 Sputum (if immediately available) for urgent Gram stain and culture. The usefulness of sputum tests remains debatable because of contamination by upper respiratory tract commensals. However, a single or predominant organism on a Gram stain of a fresh sample or a heavy growth on culture of purulent sputum is likely to be the organism responsible. The finding of many polymorphonuclear cells (PMN) with no bacteria in a patient who has not already received antibiotics can reliably exclude infection by most ordinary bacterial pathogens. Specimens should be obtained by deep cough and be grossly purulent. Ideally the specimen should be obtained before treatment with antimicrobials, if this does not delay administration of antibiotics, and be transported to the laboratory immediately for prompt processing to minimise the chance of missing fastidious organisms (e.g. Streptococcus pneumoniae). Acceptable specimens (in patients with normal or raised white blood cell (WBC) counts) should contain > 25 PMN per low-power field (LPF) and < 10–25 squamous epithelial cells (SEC)/LPF or > 10 PMN per SEC. These criteria should not be used for Mycobacterium and Legionella infection. Certain organisms are virtually always pathogens when recovered from respiratory secretions (Table 32.2). Patients with risk factors for tuberculosis (TB) (Table 32.3) and particularly those with cough for more than a month, other common symptoms of TB and suggestive radiographic changes should have sputum examined for acid-fast bacilli (AFB). Sputum cannot be processed for culture for anaerobes due to contamination by the endogenous anaerobic flora of the upper respiratory tract. In addition to the factors listed in Table 32.1, foul-smelling sputum, lung abscess and empyema should raise suspicion of anaerobic infection.

8 Aspiration of pleural fluid for Gram stain, culture, pH and leukocyte count – all patients with a pleural effusion > 1 cm thick on a lateral decubitus CXR.

9 Urinary Legionella antigen. This test is specific (> 95%). In patients with severe legionnaire’s disease sensitivity is 88–100% for L. pneumophilia serogroup 1 (the most commonly reported cause of Legionella infection). Thus a positive result is virtually diagnostic of Legionella infection but a negative result does not exclude it. In areas where other Legionella species are more common, this test is less helpful.

10 Urinary pneumococcal antigen has moderate sensitivity (50–80%) and high specificity (> 90%).

11 Microimmunofluorescence serology for Chlamydia pneumoniae immunoglobulin (Ig) M. A titre of 1:16 is significant.

12 Human immunodeficiency virus (HIV) serological status.

Table 32.2 Organisms which are virtually always pathogens when recovered from respiratory secretions ⁹

Legionella

Chlamydia

Tuberculosis

Influenza, para-influenza virus, respiratory syncytial virus, adenovirus, Hantavirus, severe acute respiratory syndrome (SARS), coronavirus

Stronglyloides stercoralis

Toxoplasma gondi

Pneumocystis carinii

Histoplasma capsulatum

Coccidiodes immitis

Blastomycoses dermatitidis

Cryptococcus neoformans

Table 32.3 Risk factors for pulmonary tuberculosis

Living in or originating from a developing country

Age (< 5 years, middle-aged and elderly men)

Alcoholism and/or drug addiction

Human immunodeficiency virus (HIV) infection

Diabetes mellitus

Lodging-house dwellers

Immunosuppression

Close contact with smear-positive patients

Silicosis

Poverty and/or malnutrition

Previous gastrectomy

Smoking

Other investigations should be considered in patients with risk factors for infection with unusual organisms. Bronchoalveolar lavage may be useful in immunocompromised patients, those who fail to respond to antibiotics or those in whom sputum samples cannot be obtained.¹¹

MANAGEMENT

GENERAL SUPPORTIVE MEASURES

Intravenous fluids may be required to correct dehydration and provide maintenance fluid. A general approach should be made to organ support with an emphasis on correcting hypoxia.

ANTIMICROBIAL REGIMES

Each unit should have its own regimens tailored to the local flora and antibiotic resistance patterns. In the absence of such regimens the regimen outlined in Figure 32.1 may be helpful. This should be modified in the light of risk factors (see Table 32.1). Quinolones may be less appropriate in areas with a high prevalence of TB as their use may mask concurrent TB infection. Appropriate antimicrobial therapy should be administered within 1 hour of diagnosis.^8,¹⁴ There is controversy regarding the appropriate change to empiric therapy based on microbiological findings.^7,⁸ Changing to narrower-spectrum antimicrobial cover may result in inadequate treatment of the 5–38% of patients with polymicrobial infection. Furthermore, dual therapy may be more effective than monotherapy, even when the identified pathogen is sensitive to the agent chosen, particularly in severely ill patients with bacteraemic pneumococcal pneumonia.¹⁵ For the treatment of drug-resistant Streptococcus pneumoniae, the regimes in Figure 32.1 are probably suitable for isolates with a penicillin minimum inhibitory concentration (MIC) < 4 mg/l.⁷ If the MIC is ≥ 4 mg/l an antipneumococcal fluoroquinolone, vancomycin, teicoplanin or linezolid should be given.⁸

Figure 32.1 Antibiotic regimes for treatment of severe community-acquired pneumonia in critically ill patients.^7,⁸ Respiratory fluoroquinolones include moxifloxacin and levofloxacin. Advanced macrolides include azithromycin and clarithromycin. Cefotaxime is a suitable non-antipseudomonal third-generation cephalosporin.

The role of zanamivir and oseltamivir in severe influenza pneumonia is not clear but early treatment of patients with less severe symptoms results in a reduction of the duration of symptoms if treatment is started early (< 48 hours from onset).¹⁶ Oseltamivir is recommended as first-line therapy for patients with suspected avian influenza A/H5N1.¹⁷

DURATION OF THERAPY

There are no clinical trials that have specifically addressed this issue. Courses as short as 5 days may be sufficient ¹⁸ but antibiotics should be continued until the patient has been afebrile for 48–72 hours and organ dysfunction has largely resolved.⁷ Short courses may be suboptimal for patients with bacteraemic Staphylococcus aureus pneumonia, meningitis or endocarditis complicating pneumonia or infection with less common organisms (e.g. Burkholderia pseudomallei or fungi) or Pseudomonas aeruginosa.

RESPONSE TO THERAPY ^7,^9,¹⁹

This can be assessed subjectively (a response is usually seen within 1–3 days of starting therapy) or objectively on the basis of respiratory symptoms, fever, oxygenation, WBC count, bacteriology and CXR changes. The average time to defevescence varies with organism, severity and patient age (7 days in elderly patients, 2.5 days in young patients with pneumococcal pneumonia, 6–7 days in bacteraemic patients with pneumococcal pneumonia, 1–2 days in patients with M. pneumoniae pneumonia and 5 days in patients with Legionella pneumonia). Both blood and sputum cultures are usually negative within 24–48 hours of treatment, although P. aeruginosa and M. pneumoniae may persist in the sputum despite effective therapy. CXR changes lag behind clinical changes with the speed of change depending on the organism, the age of the patient and the presence and absence of comorbid illnesses. The CXR of most young or middle-aged patients with bacteraemic pneumococcal pneumonia is clear by 4 weeks but resolution is slower in elderly patients and patients with underlying illness, extensive pneumonia on presentation or L. pneumophilia pneumonia.

If the patient fails to respond, consider the following questions:

• Does the patient have pneumonia?

• Are there host factors which explain the failure (e.g. obstruction of bronchus by a foreign body or tumour, inadequate host response)?

• Has a complication developed (e.g. empyema, superinfection, bronchiolitis obliterans organising pneumonia)?

• Is the right drug being given in an adequate dose by the right route?

• Is the organism resistant to the drug being given?

• Are there other organisms?

• Is the fever a drug fever?

Useful investigations include computed tomography of the chest, bronchoalveolar lavage (Table 32.4) and transbronchial or open-lung biopsy.

Table 32.4 Procedure for obtaining microbiological samples using bronchoscopy and protected specimen brushing (PSB) and/or bronchoalveolar lavage (BAL)^12,¹³

Infection control

In patients suspected of having a disease which is transmitted by the airborne route (e.g. tuberculosis):

• the risk of transmission should be carefully weighed against the benefits of bronchoscopy, which may generate large numbers of airborne particles

• perform bronchoscopy in a negative-pressure isolation room

• consider the use of a muscle relaxant in ventilated patients, to prevent coughing

• staff should wear personal protective equipment which should include a fit-tested negative-pressure respirator (N95, FFP2 or above) as a minimum. Use of a powered air-purifying respirator should be considered

General recommendations Suction through the endotracheal tube should be performed before bronchoscopy Avoid suction or injection through the working channel of the bronchoscope Perform protected specimen brushing before bronchoalveolar lavage Ventilated patients Set FiO₂ at 1.0 Set peak pressure alarm at a level that allows adequate ventilation Titrate ventilator settings against exhaled tidal volume Consider neuromuscular blockade in addition to sedation in patients at high risk of complications who are undergoing prolonged bronchoscopy Protected specimen brushing Sample the consolidated segment of lung at subsegmental level If purulent secretions are not seen, advance the brush until it can no longer be seen but avoid wedging it in a peripheral position Move brush back and forth and rotate it several times Bronchoalveolar lavage Wedge tip of bronchoscope into a subsegment of the consolidated segment of lung Inject, aspirate and collect 20 ml of sterile isotonic saline. Do not use this sample for quantitative microbiology or identification of intracellular organisms. It can be used for other microbiological analysis Inject, aspirate and collect additional aliquots of 20–60 ml The total volume of saline injected should be 60–200 ml Complications Hypoxaemia (possibly less with smaller BAL volumes) Arrhythmia Transient worsening in pulmonary infiltrates Bleeding (particularly following PSB) Fever (more common after BAL) Positive results > 5% of cells in cytocentrifuge preparations of BAL fluid contain intracellular bacteria or ≥ 10³ colony-forming units/ml in PSB specimen or ≥ 10⁴ colony-forming units/ml in BAL fluid

ADMISSION TO ICU

This will largely be determined by the need for organ support and the availability of beds. However the presence of three of the following should prompt referral for admission: systolic blood pressure ≤ 90 mmHg, multilobar disease, PaO₂/FiO₂ ratio ≤ 250 (PaO₂ in mmHg) or ≤ 33 (PaO₂ in kPa), multilobar infiltrates, confusion, urea > 7 mmol/l (20 mg/dl), leukopenia, thrombocytopenia, hypothermia and need for aggressive fluid resuscitation.⁷

HOSPITAL-ACQUIRED PNEUMONIA

Hospital-acquired pneumonia occurs in 0.5–5% of hospital patients, with a higher incidence in certain groups, e.g. postoperative patients and patients in ICU. Diagnosis may be difficult: the clinical features of pneumonia are non-specific and many non-infectious conditions (e.g. atelectasis, pulmonary embolus, aspiration, heart failure and cancer) can cause infiltrates on a CXR. Identification of the organism responsible is even more difficult than in patients with community-acquired pneumonia due to the high incidence of oropharyngeal colonisation by Gram-negative bacteria. Blood cultures are only positive in about 6% of cases of nosocomial pneumonia. Ventilator-associated pneumonia is nosocomial pneumonia arising > 48–72 hours after intubation. It is associated with a higher incidence of multidrug-resistant organisms.¹

PATHOGENESIS

Nosocomial pneumonia is thought to result from microaspiration of bacteria colonising the upper respiratory tract. Other routes of infection include macroaspiration of gastric contents, inhaled aerosols, haematogenous spread, spread from pleural space and direct inoculation from ICU personnel.

CLINICAL DIAGNOSIS

Health care-associated pneumonia is defined on the basis of time of onset (developing more than 48 hours after admission to a health care facility ¹), CXR changes (new or progressive infiltrates) and either clinical features and simple laboratory investigations or the results of quantitative microbiology. Using a clinical approach, pneumonia is diagnosed by the finding of a new infiltrate or a change in an infiltrate on CXR and growth of pathogenic organisms from sputum plus one of the following: WBC count greater than 12 × 10⁵/litre, core temperature ≥ 38.3°C, sputum Gram stain with scores of more than two on a scale of four of polymorphonuclear leukocytes and bacteria.

INVESTIGATIONS

These are broadly similar to those required in community-acquired pneumonia:

• CXR: although studies using a histological diagnosis as the gold standard have demonstrated that pneumonia may be present despite a normal CXR, most definitions of nosocomial pneumonia require the presence of new persistent infiltrates on a CXR.

• Respiratory secretions: considerable controversy surrounds the issue of whether invasive bronchoscopic sampling (see Table 32.4) of respiratory secretions is necessary. A randomised controlled trial has demonstrated a reduced 14-day mortality amongst patients with ventilator-associated pneumonia who underwent invasive bronchoscopic sampling compared to those who were treated using a non-invasive management strategy.²⁰ However the important difference between the two groups may have been the use of quantitative microbiological techniques in the former rather than the use of invasive sampling.²¹ Previous data have demonstrated a high correlation between the results of quantitative culture of invasive samples and quantitative culture of tracheal aspirates.²² A more recent trial of non-quantitative culture of bronchoalveolar lavage specimens versus non-quantitative culture of tracheal aspirates revealed no difference in any outcome measures.²³ Although tracheal aspirates may predominantly reflect the organisms colonising the upper airway they may be useful in indicating which organisms are not responsible for the pneumonia, thus allowing the antimicrobial cover to be narrowed.¹ This interpretation is based on the premise that the predominant route of infection is via the upper respiratory tract. From this it can be assumed that if the organism is not present in the upper respiratory tract the probability of it being present in the lung parenchyma is low. Certain organisms are virtually always pathogens when recovered from respiratory secretions (see Table 32.2).

• Blood cultures identify the aetiological agent in 8–20% of patients. Bacteraemia is associated with a worse prognosis. In 50% of patients with severe hospital-acquired pneumonia and positive blood cultures there is another source of sepsis.

MANAGEMENT

Management is based on the finding that early treatment with antimicrobials that cover all likely pathogens results in a reduction in morbidity and mortality.⁵ If quantitative microbiology is not used the initial selection of antimicrobials is made on the basis of epidemiological clues (Figure 32.2; Table 32.5). Antimicrobials should be administered within 1 hour of diagnosis.¹⁴ The results of microbiological investigations are used to narrow antimicrobial cover later. Treatment should be reassessed after 2–3 days or sooner if the patient deteriorates (Figure 32.3). An outline of management based on an invasive approach is given in Figure 32.4.

Figure 32.2 An outline of initial management of nosocomial pneumonia based on a non-invasive clinical approach.¹

Table 32.5 Recommended initial empiric treatment for nosocomial pneumonia ¹ *

Situation	Antibiotics
No risk factors for multidrug-resistant pathogens	Cefotaxime or
	Levofloxacin, moxifloxacin or ciprofloxacin or
	Ampicillin/sulbactam or
	Ertapenem
Antimicrobial therapy in previous 90 days or
Current hospitalisation for ≥ 5 days or	One of:
High frequency of antibiotic resistance in the specific hospital unit or	Antipseudomonal cephalosporin (cefepime or ceftazidime) or
	Antipseudomonal carbapenem (meropenem or imipenem-cilastin) or
Hospitalisation for 2 days or more in previous 90 days or	β-lactam/β-lactamase inhibitor (e.g. piperacillin-tazobactam or cefaperazone-sulbactam)
Residence in nursing home or extended-care facility or	plus one of:
Home infusion therapy (including antibiotics) or	Aminoglycoside or
Chronic dialysis within 30 days or	Antipseudomonal quinolone (levofloxacin or ciprofloxacin) plus one of the following for patients at high risk of methicillin-resistant Staphylococcus aureus (MRSA) infection:
Home wound care or
Family member with multidrug-resistant pathogen or
Immunosuppression or	Linezolid or vancomycin or teicoplanin
Bronchiectasis

* The use of dual therapy is not well supported by evidence but it does reduce the probability that the pathogen is resistant to the drugs being given. If an extended-spectrum β-lactamase-producing strain or an Acinetobacter sp. is suspected, a carbapenem should be given. If Legionella pneumophilia is suspected, use a quinolone. Risk factors for MRSA infection in areas with a high incidence of MRSA include diabetes mellitus, head trauma, coma and renal failure.

Figure 32.3 Subsequent management of nosocomial pneumonia based on a non-invasive clinical approach.¹

Figure 32.4 Management of suspected nosocomial pneumonia based on invasive sampling of respiratory secretions.

DURATION OF THERAPY

Current ATS guidelines recommend 7 days’ treatment provided the aetiological agent is not P. aeruginosa and the patient has a good clinical response with resolution of clinical features of infection.¹ The outcome of patients who receive appropriate initial empiric therapy for ventilator-associated pneumonia for 8 days is similar to those who receive treatment for 14 days.²⁴

RESPONSE TO THERAPY

Clinical improvement is usually not apparent for 48–72 hours and therapy should not be changed in this time. The CXR is of limited value for assessing response: initial deterioration is common and improvement often lags behind clinical response. However a rapidly deteriorating CXR pattern with a > 50% increase in size of infiltrate in 48 hours, new cavitation or a significant new pleural effusion should raise concern. If the patient fails to respond consider the diagnosis, host factors (e.g. immunosuppressed, debilitated), bacterial factors (e.g. virulent organism) and therapeutic factors (e.g. wrong drug, inadequate dose). Review the antibiotics and repeat cultures. It may be useful to broaden the antimicrobial cover while waiting for the results of investigations. Consider invasive sampling of respiratory secretions, computed tomography or ultrasound of the chest (to look for an empyema or abscess), another source of infection, open-lung biopsy to establish diagnosis and aetiology, or administration of steroids.

PREVENTION

A number of measures have been shown to reduce the incidence of ventilator-associated pneumonia.²⁵^–³¹ Measures recommended by the Centers for Disease Control (CDC) include hand-washing, nursing patients in a 30° head-up position, subglottic aspiration of secretions, orotracheal rather than nasotracheal intubation, changing the breathing circuit only when visibly soiled or mechanically malfunctioning and preferential use of non-invasive ventilation. CDC guidelines can be accessed via the link page: http://www.aic.cuhk.edu.hk/web8/Pneumoniaguidelines.htm.

TUBERCULOSIS

The main risk factors are listed in Table 32.3. Typical clinical features include fever, sweating, weight loss, lassitude, anorexia, cough productive of mucoid or purulent sputum, haemoptysis, chest wall pain, dyspnoea, localised wheeze and apical crackles. Patients may also present with unresolved pneumonia, pleural effusions, spontaneous pneumothorax and hoarseness or with enlarged cervical nodes or other manifestations of extrapulmonary disease. Clinical disease is seldom found in asymptomatic individuals, even those with strongly positive tuberculin test (Heaf grade III or IV). Older patients, who may have coexistent chronic bronchitis, can be missed unless a CXR is taken. The outlook for patients with TB who require ICU admission is poor. In one retrospective study the in-hospital mortality for all patients with TB requiring ICU admission was 67% but in those with acute respiratory failure it rose to 81%.³² The presentation and management of TB in HIV-positive patients is different (see below).

INVESTIGATION OF PULMONARY TUBERCULOSIS

IDENTIFICATION OF MYCOBACTERIA ³³

Multiple (3–6) sputum samples should be collected, preferably on different days, for microscopy for AFB and culture. If sputum is not available bronchial washings taken at bronchoscopy and gastric lavage or aspirate samples should be obtained. Gastric aspirates need to be neutralised immediately on collection. Bronchoscopy and transbronchial biopsy may be useful in patients with suspected TB but negative sputum smear. Pleural biopsy is often helpful and mediastinoscopy is occasionally needed in patients with mediastinal lymphadenopathy. Part of any biopsy specimen should always be sent for culture. Nucleic acid amplification tests on sputum have a sensitivity similar to culture in smear-negative patients with pulmonary TB but have the advantage of a much more rapid result. There is, however, a significant false-negative rate.³⁴

CHEST X-RAY

A normal CXR almost excludes TB (except in HIV-infected patients) but endobronchial lesions may not be apparent and early apical lesions can be missed. Common appearances include patchy/nodular shadowing in the upper zones (often bilateral), cavitation, calcification, hilar or mediastinal lymphadenopathy (may cause segmental or lobar collapse), pleural effusion, tuberculomas (dense round or oval shadows) and diffuse fine nodular shadowing throughout the lung fields in miliary TB. Inactivity of disease cannot be inferred from the CXR alone. This requires three negative sputum samples and failure of any lesion seen on CXR to progress. CXR appearances in HIV-positive patients with TB differ from non-HIV-infected patients.

TREATMENT OF PULMONARY TUBERCULOSIS ^34,³⁵

The most commonly used regimen consists of a 6-month course of rifampicin 600 mg/day (450 mg for patients < 50 kg) and isoniazid 300 mg/day plus a 2-month course of pyrazinamide 2 g/day (1.5 g for patients < 50 kg) and ethambutol 15 mg/kg daily (streptomycin can be substituted for ethambutol). Ethambutol should only be used in patients who have reasonable visual acuity and who are able to appreciate and report visual disturbances. Visual acuity and colour perception must be assessed (if ethambutol is to be used) and liver and renal function checked before treatment is started. Steroids are recommended for children with endobronchial disease and, possibly, for patients with tuberculous pleural effusions. Pyrazinamide 10 mg/day should be given to prevent isoniazid-induced neuropathy to those at increased risk (e.g. patients with diabetes mellitus, chronic renal failure or malnutrition or alcoholic or HIV-positive patients).

INFECTION CONTROL

Patients admitted to an ICU with infectious TB or suspected of having active pulmonary TB should be managed in an isolation room with special ventilation characteristics, including negative pressure. Patients should be considered infectious if they are coughing or undergoing cough-inducing procedures or if they have positive AFB smears and they are not on or have just started chemotherapy or have a poor clinical or bacteriologic response to chemotherapy.^13,³⁴ Patients with non-drug-resistant TB should be non-infectious after 2 weeks of treatment, which includes rifampicin and isoniazid.³⁴ As TB is spread through aerosols it is probably appropriate to isolate patients who are intubated even if only their bronchial washings are smear-positive. Staff caring for patients who are smear-positive should wear personal protective equipment including a fit-tested negative-pressure respirator (N95, FFP2 or higher). Use of a powered air-purifying respirator should be considered when bronchoscopy is being performed.¹³ Detailed infection control advice can be obtained via the link page http://www.aic.cuhk.edu.hk/web8/Pneumoniaguidelines.htm-.

PNEUMONIA IN THE IMMUNOCOMPROMISED

The lungs are amongst the most frequent target organs for infectious complications in the immunocompromised. The incidence of pneumonia is highest amongst patients with haematological malignancies, bone marrow transplant recipients and patients with acquired immunodeficiency syndrome (AIDS).

The speed of progression of pneumonia, the CXR changes (Table 32.6) and the type of immune defect provide clues to the aetiology. Bacterial pneumonias progress rapidly (1–2 days) whereas fungal and protozoal pneumonias are less fulminant (several days to a week or more). Viral pneumonias are not usually fulminant but on occasions may develop quite rapidly. Bronchoscopy is a major component of the investigation of these patients. Empiric management based on CXR appearances is outlined in Table 32.6. There is some evidence that early non-invasive ventilation may improve outcome amongst immunocompromised patients with fever and bilateral infiltrates.³⁶

Table 32.6 Causes of chest X-ray changes and empiric treatment of pneumonia in the immunocompromised

Chest X-ray appearance	Causes	Empiric treatment for suspected pneumonia
Diffuse infiltrate	Cytomegalovirus and other herpesviruses	Broad-spectrum antibiotics for at least 48 hours (e.g. third-generation cephalosporin and aminoglycoside)
	Pneumocystis carinii
	Bacteria
	Aspergillus (advanced)	Co-trimoxazole
	Cryptococcus (uncommon)	Lung biopsy or lavage within 48 hours or full 2-week course of co-trimoxazole (depends on patient tolerance of invasive procedure)
	Non-infectious causes, e.g. drug reaction, non-specific interstitial pneumonitis, radiation pneumonitis (uncommon), malignancy, leukoagglutinin reaction
Focal infiltrate	Gram-negative rods	Broad-spectrum antibiotics
	Staphylococcus aureus	If response seen, continue treatment for 2 weeks
	Aspergillus	If response seen, continue treatment for 2 weeks
	Cryptococcus	If disease progresses, lung biopsy/aspirate within 48–72 hours or empiric trial of antifungal ± macrolide
	Nocardia
	Mucor
	Pneumocystis carinii (uncommon)
	Tuberculosis
	Legionella
	Non-infectious casues (e.g. malignancy, non-specific interstitial pneumonitis, radiation pneumonitis)

PNEUMOCYSTIS JIROVECI PNEUMONIA (PCP)³⁷

The incidence of this common opportunistic infection has fallen substantially in patients with AIDS who are receiving prophylaxis and effective antiretroviral therapy, with most cases occurring in patients who are not receiving HIV care or among patients with advanced immunosuppression. The onset is usually insidious with dry cough, dyspnoea and fever on a background of fatigue and weight loss. Crackles in the chest are rare. Approximately 15% of patients have a concurrent cause for respiratory failure (e.g. Kaposi sarcoma, TB, bacterial pneumonia). Useful investigations are:

1 CXR: classical appearance is diffuse bilateral perihilar interstitial shadowing but in the early stages this is very subtle and easily missed. The initial CXR is normal in 10%. In a further 10% the changes are atypical with focal consolidation or coarse patchy shadowing. None of the changes are specific for PCP and may be seen in other lung diseases asssociated with AIDS. Pleural effusions, hilar or mediastinal lymphadenopathy are unusual in PCP but common in mycobacterial infection or Kaposi’s sarcoma or lymphoma.

2 Induced sputum: in this technique the patient inhales nebulised hypertonic saline from an ultrasonic nebuliser. This provokes bronchorrhoea and the patient coughs up material containing cysts and trophozoites. The technique is time-consuming, requires meticulous technique and is less sensitive than bronchoscopy but less invasive. The possibility of concurrent TB should be considered and steps taken to minimise the risk of spread of infection.

3 Bronchoscopy with bronchoalveolar lavage leads to the diagnosis in over 90% of cases. Specimens should be sent for cytology. Transbronchial biopsy is not necessary in most cases.

Treatment should be started as soon as the diagnosis is suspected. Although there is some suggestion of potential benefit for early retroviral therapy for patients with HIV and PCP, some centres delay initiation of antiretrovirals until completion of PCP treatment in order to reduce the risk of an immune reconstitution syndrome. Treatment of choice is trimethoprim plus sulphamethoxazole (co-trimoxazole) 20 mg/kg per day + 100 mg/kg per day for 3 weeks plus prednisolone 40 mg orally twice daily for 5 days followed by 20 mg twice daily for 5 days and then 20 mg/day until the end of PCP treatment. Side-effects of co-trimoxazole are common in HIV patients (nausea, vomiting, skin rash, myelotoxicity). The dose should be reduced by 25% if the WBC count falls. Patients who are intolerant of co-trimoxazole should be treated with:

• pentamidine 4 mg/kg per day IV or

• primaquine with clindamycin or

• trimetrexate with leucovorin (± oral dapsone)

Response to treatment is usually excellent, with a response time of 4–7 days. If the patient deteriorates or fails to improve: consider (re-) bronchoscopy (is the diagnosis correct?), treat co-pathogens and consider a short course of high-dose intravenous methylprednisolone and/or diuretics (patients are often fluid-overloaded). Approximately 40% of patients with HIV-related PCP who require mechanical ventilation survive to hospital discharge.³⁸

BACTERIAL PNEUMONIA ³⁷

This is the most common cause of acute respiratory failure in HIV-positive patients. Bacterial pneumonia is more common in HIV-infected patients than in the general population and tends to be more severe. Streptococcus pneumoniae, Haemophilus influenzae, Pseudomonas aeruginosa and Staphylococcus aureus are the commonest organisms. Nocardia and Gram-negatives should also be considered. Atypical pathogens (e.g. Legionella) are rare. Response to appropriate antibiotics is usually good but may require protracted courses of antibiotics because of high tendency to relapse. Patients with severe immunodeficiency (CD4⁺ T-lymphocyte count < 100/μl) and a history of Pseudomonas infection or bronchiectasis or neutropenia should receive antibiotics that cover P. aeruginosa as well as other Gram-negatives. The possibility of concurrent PCP or TB should be excluded.

TUBERCULOSIS

TB may be the initial presentation of AIDS, particularly in sub-Saharan Africa. The pattern of TB in HIV patients depends on the degree of immunosuppression. In patients with CD4⁺ T lymphocytes > 350 cells/μl the clinical presentation is similar to TB in non-HIV-infected patients, although extrapulmonary disease is more common. In patients with CD4⁺ T lymphocytes < 350 cells/μl extrapulmonary disease (pleuritis, pericarditis, meningitis) is common. Severely immunocompromised patients (CD4⁺ T lymphocytes < 100 cells/μl) may present with severe systemic disease with high fever, rapid progression and systemic sepsis. In these patients lower and middle-lobe disease is more common, miliary disease is common and cavitation is less common. Sputum smears and culture may be positive even with a normal CXR.

Response to treatment is usually rapid. Complex interactions occur between rifamycins (e.g. rifampicin and rifabutin) and protease inhibitors and non-nucleosidase reverse transcriptase inhibitors used to treat patients infected with HIV. The choice of rifampicin or rifabutin depends on a number of factors, including the unique and synergistic adverse effects for each individual combination of rifampicin and anti-HIV drugs and consultation with a physician with experience in treating both TB and HIV is advised.³⁹ Infectious Diseases Society of America-recommended dosage adjustment for patients receiving antiretrovirals and rifabutin³⁷ can be obtained via the link page: http://www.aic.cuhk.edu.hk/web8/Pneumoniaguidelines.htm . The optimal time for initiating antiretroviral therapy in patients with TB is controversial. Early therapy may decrease HIV disease progression but may be associated with a high incidence of adverse effects and an immune reconstitution reaction.³⁷

CYTOMEGALOVIRUS (CMV) PNEUMONITIS ^40,⁴¹

Risk of infection is highest following allogeneic stem cell transplantation, followed by lung transplantation, pancreas transplantation and then liver, heart and renal transplantation and advanced AIDS. If both the recipient and the donor are seronegative then the risk of both infection and disease is negligible. If the recipient is seropositive the risk of infection is approximately 70% but the risk of disease is only 20%, regardless of the serostatus of the donor. However, if the recipient is seronegative and the donor is seropositive, the risk of disease is 70%. If steroid pulses and antilymphocyte globulin are given for treatment of acute rejection the risk of developing disease is markedly increased. Infection may be the result of primary infection or reactivation of latent infection. It is clinically important, but often difficult, to distinguish between CMV infection and CMV disease and a definitive diagnosis can only be made histologically. Detection of CMV-pp65 antigen in peripheral WBCs and detection of CMV DNA or RNA in the blood by quantitative polymerase chain reaction are the most useful tests for demonstrating CMV disease. Using thresholds of 10/300 000–50/200 000 positive circulating peripheral WBC, the positive predictive value for CMV-pp65 ranges from 64% to 82% and the negative predictive value from 70% to 95%.⁴²^–⁴⁴ Treatment consists of intravenous ganciclovir for at least 14 days. Foscarnet can be used if ganciclovir fails.

FUNGAL PNEUMONIA

Fungi are rare but important causes of pneumonia. They can be divided into two main groups based on the immune response required to combat infection with these organisms. Histoplasma, blastomycosis, coccidioidomycosis, paracoccidioidomycosis and Cryptococcus require specific cell-mediated immunity for their control and thus, in contrast to infections which are controlled by phagocytic activity, the diseases caused by these organisms can occur in otherwise healthy individuals, although they cause much more severe illnesss in patients with impaired cell-mediated immunity (e.g. patients infected with HIV and organ transplant recipients). With the exception of Cryptococcus, these organisms are rarely seen outside North America. Aspergillus and Mucor spores are killed by non-immune phagocytes and as a result these fungi rarely result in clinical illness in patients with normal neutrophil numbers and function.

CANDIDIASIS

This is effectively a combination of the two types of fungal infection in that impaired cell-mediated immunity predisposes to mucosal overgrowth with Candida but impaired phagocytic function or numbers is usually required before deep invasion of tissues occurs. Primary Candida pneumonia (i.e. isolated lung infection) is uncommon ^40,^45,⁴⁶ and more commonly pulmonary lesions are only one manifestation of disseminated candidiasis. Even more common is benign colonisation of the airway with Candida. In most reported cases of primary Candida pneumonia amphotericin B has been used. In disseminated candidiasis treatment should be directed to treatment of disseminated disease rather than Candida pneumonia per se.⁴⁶

INVASIVE ASPERGILLOSIS ⁴⁷

This is a highly lethal condition in the immunocompromised despite treatment and therefore investigation and treatment should be prompt and aggressive. Definitive diagnosis requires both histological evidence of acute-angle branching, septated nonpigmented hyphae measuring 2–4 μm in width and cultures yielding Aspergillus species from biopsy specimens of involved organs. Recovery of Aspergillus species from respiratory secretions in immunocompromised, but not immunocompetent, patients may indicate invasive disease with a positive predictive value as high as 80–90% in patients with leukaemia or bone marrow transplant recipients. Bronchoalveolar lavage with smear, culture and antigen detection has excellent specificity and reasonably good positive predictive value for invasive aspergillosis in immunocompromised patients. Although radiological features may give a clue to the diagnosis they are not sufficiently specific to be diagnostic. Characteristic CXR features (wedge-shaped pleural-based densities or cavities) occur late. The ‘halo sign’ (area of low attenuation surrounding a nodular lung lesion) is an early computed tomography finding whereas the ‘crescent sign’ (an air crescent near the periphery of a lung nodule) is a late feature.

In acutely ill immunocompromised patients intravenous therapy should be initiated if there is suggestive evidence of invasive aspergillosis while further investigations to confirm or refute the diagnosis are carried out. First-line therapy is voriaconazole.⁴⁸ Caspofungin and amphotericin are alternatives.

PARAPNEUMONIC EFFUSION

This may be an uncomplicated effusion which resolves with appropriate treatment of the underlying pneumonia or a complicated effusion which develops into an empyema unless drained. Complicated effusions tend to develop 7–14 days after initial fluid formation. They are characterised by increasing pleural fluid volume, continued fever and pleural fluid of low pH (< 7.3) which contains a large number of neutrophils and may reveal organisms on Gram staining or culture. An outline of management is given in Figure 32.5.

Figure 32.5 An approach to the management of parapneumonic effusions.

An empyema arises following infection of the structures surrounding the pleural space, including subdiaphragmatic structures, and chest trauma or may be associated with malignancy. Anaerobic bacteria, usually streptococci or Gram-negative rods, are responsible for 76% of cases.

DIAGNOSIS

The diagnosis is simple. The patient is usually septic and may have a productive cough and chest pain. The CXR may show features suggestive of a pleural effusion and underlying consolidation but may also show an abscess cavity with a fluid level, in which case computed tomography scanning will be required to distinguish between an abscess and an empyema. Ultrasound can be useful to confirm the presence of fluid in the pleural space and to determine whether it can be drained by needle aspiration or, if there is debris within the fluid, drainage using an intercostal drain. The diagnosis is confirmed by aspiration of pus.

TREATMENT

The mainstay of treatment is drainage either by intercostal drain or by surgical intervention. Patients who present before the pus is loculated and a fibrinous peel has formed on the lung can usually be treated by simple drainage. The combination with intrapleural fibrinolysis may be beneficial.⁵⁰ Optimal surgical management, which consists of decortication (open or thoracoscopic), is indicated if the empyema is more advanced or if simple drainage fails. This is a major procedure and many patients with cardiac or chronic respiratory disease will not tolerate it. Alternatives for these patients are instillation of thrombolytics into the pleural space or thoracostomy. Antibiotics have only an adjunctive role. Broad-spectrum antibiotic regimes with anaerobic cover should be used until the results of microbiological analysis of the aspirated pus are available.