Background The majority of childhood community acquired pneumonia (CAP) in developing countries is believed to be bacterial in origin. However, the predictors of bacterial versus non-bacterial (viral) pneumonia are not clearly defined. This is essential for judicious use of antibiotic therapy.

Objective To determine the microbiologic aetiology of childhood CAP in India, and determine the predictors of bacterial pneumonia.

Methods Children (1 month^-12 years) fulfilling World Health Organisation criteria for pneumonia (cough or difficult breathing, and tachypnea; for <7 days) were enrolled through a two-year (April²011-March2013) surveillance programme. Pneumonia severity was assessed using WHO criteria. Nasopharyngeal aspirate (NPA) culture, blood culture, IgM anti-Mycoplasma pneumoniae and IgM anti-Chlamydia pneumoniae were examined. Demographic characteristics, clinical profile, presence of ‘risk factors’, clinical examination findings, and radiographic features were evaluated as predictors of bacterial aetiology.

Results 2333 children with CAP were enrolled. 61% were 5–12 years. Figure 1 presents the pneumonia severity. Bacterial pathogens were isolated in 12.7% NPA cultures with Pneumococcus (n = 223), Staphylococcus aureus (n = 27), and Haemo philus influenzae (n = 23) predominating. Blood culture yielded bacteria in only 3.3%. S. aureus (n = 25), Gram negative bacilli (n = 21), and Alpha-hemolytic Streptococcus (n = 15) predominated. Pneumococcus (n = 3) accounted for a minority. Serology for Mycoplasma and Chlamydia were positive in 4.4% and 1.6% samples respectively (Figure 2A,2B).

Table 1 highlights the unadjusted odds ratio for various factors explored as predictors of bacterial aetiology. Exposure to over-crowding at home appeared to be associated with a lower risk of bacterial aetiology, whereas exposure to tobacco smoke was associated with higher risk. None of the other factors predicted bacterial aetiology.

Conclusion The majority of childhood community acquired pneumonia appears to be non-bacterial in origin. Bacterial aetiology could not be predicted by demographic, clinical, or radiographic features, that are usually believed to be associated with bacterial aetiology.