Alternatives for detecting the pathogen: clinical, radiological, biochemical features

Bacterial and viral pneumonia cannot be differentiated reliably on clinical or radiological grounds,2 ,3 ,5 although the presence of pleural effusions suggests pneumococcal infection.6 The clinical features of CAP (fever, tachypnoea, breathlessness or difficulty in breathing, cough, wheeze, chest pain) vary and tend not to be specific for bacterial or viral infections. Mycoplasma pneumonia is the possible exception, where chest pain is frequently noted.7 Aetiology is difficult to assign radiologically as mild bacterial infections often produce less marked changes on the chest radiograph and conversely, some severe viral infections produce lobar changes.5 ,8

The use of biomarkers, either on their own or within clinical prediction rules, has been explored. In children with fever, C-reactive protein (CRP) and procalcitonin (PCT) performed better than interferons or erythrocyte sedimentation rate at ruling-in rather than ruling-out bacterial infections.9 High levels of PCT (>1.5 ng/mL) and CRP (>100 mg/L) have been strongly associated with pneumococcal CAP, with a sensitivity of 94.4% for PCT and 91.9% for CRP.10 Although PCT is a promising biomarker in adults with pneumonia, its diagnostic threshold in childhood pneumonia is less defined and its usefulness and safety in guiding management have not been established. If levels of PCT and CRP are very high, then bacterial pneumonia is likely; however, positive and negative predictive values are not yet discriminating enough to rely on for treatment decisions.9 ,11 In the future, increasingly sophisticated clinical prediction rules incorporating biomarkers and clinical features may be usefully incorporated as a clinical decision-making tool.

Pathogen detection: what tests can be done?

How should samples be taken?

Correct samples are needed to make a definitive microbiological diagnosis. Easily accessible samples include blood, urine, nose and throat swabs and nasopharyngeal aspirates (NPAs). In children, it is often difficult to obtain adequate sputum specimens, but when possible they are also useful. Additionally, the utility of induced sputum is currently being explored.19 Samples requiring invasive procedures including bronchoalveolar lavage and lung aspiration are infrequently done and may require sedation or a general anaesthetic in children.

It is important to recognise that the method and site of respiratory sampling can impact on yield. For example, the overall sensitivity of viral detection with NPA samples is higher than that obtained with nasal swab (NS) samples. NS is an inadequate sample if IF is used for viral detection, but if PCR methods are used, diagnostic yields are comparable except for RSV.20 Adenovirus may be more frequently detected on NPA samples than nasal washes, whereas the reverse may be true for coronavirus.21

Which tests can be performed on respiratory secretions?

A century ago, sputum and lung aspirates were obtained routinely and microbiological methods were refined, resulting in an extremely high identification of causative pathogens.22 A failure of diagnostic testing is not just isolated to children. In his review, Bartlett displays sequential studies showing that the yield of pneumococcal identification in adults dropped from 81% in the 1930s to 12% in 2003 and 7.6% in 2009. This does not reflect a dramatic decline in pneumococcal infection but rather the lack of diagnostic microbiology performed as well as longer time to present and lower use of presampling antibiotics in historical cases. A major component of diagnosis was previously lung aspiration. Lung aspirates in children increase the diagnostic yield, successfully identifying pathogens in approximately 60% of children, and are still performed in selected populations.23 However, given lung aspirates have a small risk of serious adverse events and most children with pneumonia recover uneventfully, these are generally performed clinically only where the diagnostic benefits outweigh the risks.

Though nasopharyngeal secretions, nose swabs and throat swabs are frequently and easily obtained in children and multiplexed respiratory viral PCRs are routinely available, allowing rapid viral identification, bacterial PCRs remain difficult to interpret. Their diagnostic discrimination is limited by their ability to determine infection from colonisation or prolonged shedding, especially for S pneumoniae in young children. This may also be true for Mycoplasma PCR on NPAs24 and, although not well explored, viral carriage and prolonged shedding are also a clinical possibility.25 As molecular diagnostics improve and become more sensitive this issue becomes more relevant.12 There is some promising evidence to suggest that quantitative pneumococcal PCRs may be able to differentiate infection from colonisation;26 ,27 however, at present bacterial PCRs when applied to upper airway secretions not clinically useful in CAP.

Which tests can be performed on blood samples?

Blood culture is influenced by prior antibiotics and requires a high bacterial load for successful culture. Most children are not bacteraemic even with bacterial pneumonia,2 ,28 ,29 although in a series of children hospitalised mostly with lobar pneumonia or effusions, 9% had positive blood cultures.16 When positive, cultures are usually S pneumoniae, but occassionally Staphylococcus aureus especially in complicated pneumonia.30

Except for M pneumoniae, serology in most cases is impractical, given the need for acute and convalescent samples seldom achieved. Group A streptococcal infection (GAS) is also usefully diagnosed by serology, and though important in terms of severity, this infection is frequently overlooked, despite often progressing to intensive care admission or empyema31 and being found in up to 10% of hospitalised CAP.28

Pneumococcal antibody and antigen testing on blood are not routinely available in most centres and remain useful mainly in the context of epidemiological research. Both have been usurped in the clinical arena with the advent of PCR.

There is accumulating evidence for the use and validity of pneumococcal RT PCR in children with pneumonia. The application of RT PCR on blood, targeting either the ply gene32 or lytA gene17 improves pneumococcal detection between 232 and 516 ,17 times. Reassuringly, RT PCR using lytA as the target gene, appears specific, and is not detected in healthy carriers.33 Quantitative RT PCR may be predictive of pneumonia severity as a few studies have demonstrated that a high bacteraemic DNA load was associated with increased mortality.34

Which tests can be performed on pleural fluid?

Pleural fluid is an excellent specimen for microbiological identification, although it requires an invasive procedure to obtain, and thus usually a general anaesthetic in children. Gram stain of pleural fluid has a good positive predictive value but is insensitive.12 Applying non-culture techniques to pleural fluid increases diagnostic yields significantly. Antigen detection using latex agglutination and ICTs as well as PCR have been described. Latex agglutination has been variously successful, with positive results between 25% and 90%.35 ,36 Binax NOW is sensitive (71%–96%) and specific (71%–100%),12 though performs less well than pleural fluid PCR does.36

Pneumococcal PCRs have been applied to pleural fluid in many studies, with up to 75% infections identified as pneumococcal.35 ,37 Pleural fluid pneumococcal serotyping by either PCR16 ,32 ,37 or multiplex IAs for capsular polysaccharides35 often identifies serotype 1 as predominant.

Multiplex RT PCRs have successfully detected S pneumoniae and S aureus (including methicillin-resistant), S pyogenes, Haemophilus influenzae (HI) and M pneumoniae in 84% of pleural fluid; significantly better than by culture.37 A 16S ribosomal DNA PCR has also improved bacterial detection, in one study identifying an additional 28% of pathogens, mostly pneumococcal but also GAS, S aureus and haemophilus influenzae (HI).36 However, when compared with species specific PCR, 16S PCR appears less sensitive.18

Which tests can be performed on urine?

Urine is easily obtainable and the Binax NOW S pneumoniae assay is recommended in diagnostic testing in adults.12 In children, there have been concerns that urinary antigen testing is confounded by pneumococcal carriage, vaccination and prolonged excretion of pneumococci.12 ,13 It may, however, be more specific for disease in symptomatic children and has recently been suggested that it may be a useful diagnostic adjunct in older children.10 ,38