Globally, respiratory diseases are a leading cause of morbidity and mortality among both children and adults 12. In developing countries, children less than 5 years of age are at high risk for severe, life-threatening disease associated with bacterial and viral pathogens 3. Streptococcus pneumoniae is a major respiratory pathogen, and the spectrum of clinical presentations highly associated with this pathogen includes bacteremic and non-bacteremic presentations of pneumonia as well as parapneumonic effusions or empyema 45.

Several studies from developed countries suggest that the prevalence of empyema and pleural effusion may be increasing 678910. In these countries, pediatric empyema is often quickly identified and treated promptly with surgical intervention or pharmacologic therapy 1112. Nevertheless, empyema is associated with prolonged hospitalization stays (mean ~7 days) and with a case-fatality rate of about 5–7% 13. While predictors of empyema in hospitalized children are not well-known, it appears that host factors may play a predisposing role 14. In developing countries, severe pneumonia in children may be associated with necrotizing changes in a unilateral or bilateral pattern 1516. In developing countries, antecedent conditions such as malnutrition, measles or infection with antibiotic-resistant organisms may increase the risk of severe pneumonia accompanied by empyema 1718. In South Korea, the 7-valent conjugate pneumococcal vaccine (PCV7) was licensed in 2002 and coverage has reached to ~30% in the infant age group 19. In Taiwan, PCV7 was introduced in 2006 and coverage is <10%. In both Korea and Taiwan, pneumococcal polysaccharide vaccines (PPV) are available but uptake has been low in older children and adults 20. In China and Vietnam, the PCV7 has not been licensed (likely to occur in next 2 years) and uptake of PPV has also been slow and coverage is low (~1%).

The mean age of children with empyema and pleural effusion in developed country studies is 3–6 years with 50% to 80% of cases occurring in males 212223. In previous studies, bacterial pathogens that included Staphylococcus aureus, S. pneumoniae, Streptococcus pyogenes and Haemophilus influenzae type b (Hib) were associated with empyema in children 2425. However, an emerging body of literature now suggests that S. pneumoniae in particular is a major cause of empyema and that selected serotypes of pneumococcus may play an important role in this emerging disease pattern 10.

In the Asia-Pacific region, a limited number of clinical and laboratory studies suggest that S. pneumoniae may also be the most common etiologic agent in empyema and pleural effusion specimens 262728. In order to better understand epidemiologic and microbiologic patterns of empyema and pleural effusion among diverse populations of Asian children <16 years of age, we undertook a retrospective review of hospital records in four countries.

Our study results show that pleural effusion and empyema occur in Asia among young children at rates similar to those observed elsewhere (1–3% of pneumonia admissions) 293031. However, the clinical differentiation between pleural effusion and empyema may be interpreted differently from one country to another, as suggested by the great difference in the proportion of the two types reported in the four countries studied. For example, in China, 100% of the pleural effusion and empyema cases were coded only as pleural effusion, but it is likely that these included many empyema cases. In contrast, in Taiwan, only 4% of the cases were coded as pleural effusion, suggesting either that virtually no pleural fluid samples were taken from children diagnosed with uncomplicated pleural effusions or that all patients from whom samples are taken are automatically coded as "empyema". In Korea and Vietnam, the proportions were more mixed, but coding for pleural effusion nonetheless predominated. Regardless of the distinction, these findings emphasize the importance of looking at coding for both pleural effusion and empyema in order to understand the epidemiology of complicated pneumonias.

In this review, one-third of the empyema patients identified were less than 2 years of age confirming that empyema may be more likely to occur in young children than in older children. In India, one-third of hospitalized children with empyema were <5 years of age 26. The mean age (4~5 years) of children with empyema and pleural effusion in our study was similar to findings elsewhere 323334.

The absence of a distinct seasonality in the distribution of empyema patients in this study suggests weather may not be an important contributing factor or that the etiologic agents of empyema and pleural effusions specimens could be a heterogeneous group of infectious agents. This conclusion is supported, in part, by data from microbiologic cultures of empyema and pleural fluid showing a wide variety of Gram-positive and Gram-negative organisms as well as fungi and culture-negative specimens.

Our study identified several children whose empyema or pleural fluid cultures grew bacterial pathogens normally associated with community-acquired lower respiratory tract disease including Hib, S. pneumoniae and S. aureus 353637. These findings are consistent with a growing number of reports suggesting that much childhood empyema could be vaccine-preventable 3839. In our study hospitals, data on bacterial species also suggest either that a large proportion of children may acquire Gram-negative pathogens as nosocomial infections or that a number of bacterial isolates are contaminants of laboratory cultures 4041. Previous reviews or case series describing bacterial organisms isolated from children with empyema suggest that Gram-positive as well as Gram-negative organisms may invade the pleural space 4243.

In our study, a high proportion (71%) of all empyema and pleural fluid specimens grew no bacterial pathogen. This finding is consistent with a number of previous studies suggesting that the negative cultures are not the result of limitations in routine microbiology laboratory procedures. The negative cultures more likely are due to the widespread use of antibiotics (including inappropriately chosen or dosed antibiotics) as well the potential for severe viral lower respiratory tract disease to be associated with pleural effusion or bacterial superinfections resulting in necrotizing pneumonia and empyema 4445. This study collected data from existing computerized hospital discharge databases and laboratory records but individual patient medical records were not accessed to obtain information on prior treatment with antibiotics. In general, in our previous studies, we have found that parent- or patient-reported prior use of antibiotics is often not recorded in medical records. In addition, the high rate of negative cultures may be due to the presence of fastidious organisms such as anaerobic bacteria 46. Based on our previous studies in Asia, we have found that many hospital laboratories do not use anaerobic culture media. According to a recent study reported by Song JH et al, the prevalence of penicillin resistance in S. pneumoniae isolates was 71.4% in Vietnam followed by Korea (54.8%), Taiwan (38.6%) and China (23.4%) 47. Given the high proportion of bacterial culture-negative pleural fluid specimens, a more complete assessment of parapneumonic pleural effusions or empyema in prospective studies could apply non-culture-based antigen detection or polymerase chain reaction tests to detect both bacterial and viral pathogens. These tests have been used to identify children with culture-negative pneumococcal infections 43484950. These sensitive diagnostic tools can help us to better understand the burden of disease, trends over time, epidemiological differences among countries, patient demography, symptomatology, etiological agents and rational treatment. Unfortunately, at present, these laboratory techniques are only generally available in research laboratories.

A number of investigators have shown that laboratory testing of pleural fluid or empyema specimens in children with pneumonia can provide important insights into the origins of the pneumonia 5152. In Asia relatively few children hospitalized with serious pneumonia undergo thoracocentesis diagnostic procedures to identify pathogens in pleural or empyema fluid, which are relatively less culturally acceptable in Asian populations 53. Nevertheless, given the number of children identified in our retrospective survey, prospective multi-center studies in Asia are likely to yield substantial numbers of patients with empyema or complicated pleural effusions and shed light on the etiological agent associated with parapneumonic empyema and pleural effusions.

In this study, we found that children with empyema were significantly more likely to have positive bacterial cultures compared with children in whom pleural effusion specimens were collected. These data are consistent with previous studies suggesting that empyema fluid is the result of established infections and inflammatory reactions 2854. However, given the fact that clinicians in Asia have also found it necessary to collect clinical specimens from patients with pleural effusions, it is likely that prospective studies that include an evaluation of bacterial pathogens in pleural effusions specimens will yield a more accurate picture of the total burden of disease associated with invasive bacterial pneumonia and parapneumonic bacterial infections.

This study has some limitations. First, as a retrospective review, our data collection, analysis and reporting were restricted to that available in hospital databases or logbooks. Thus, for some years of data, incomplete patient information precluded further analysis. Computerized hospital administrative databases were accessed by collaborating study investigators. Nevertheless, in some hospitals we found that current levels of data entry limited the amount of clinical and historical data available. The increasing use of electronic medical records in Asia suggests that additional patient clinical and laboratory data are likely to be available in a number of countries in coming years. Finally, because the hospitals in our review did not record more than one ICD-10 diagnostic code, we were unable to determine other clinical conditions that the children may have had at the time of their hospitalization and thus we could not determine the proportion of children with pneumonia or lower respiratory tract infections among those who had empyema or pleural fluid specimens collected.

Similarly, as the hospital laboratories did only limited testing, there was no means to identify underlying causes of lower respiratory tract disease in infants and young children. In addition, most clinical laboratories in the study countries do not routinely preserve bacterial isolates from pleural fluid specimens because they lack resources and awareness of the utility of such specimens for research and advancement of treatment.

Future incidence studies in such hospitals to determine the true burden of parapneumonic empyema may be feasible if catchment areas for study sites can be well-defined. Recent studies from France 4255 suggest that the incidence of empyema may vary over time. Our results suggest that prospective surveillance for pneumonia with empyema or pleural effusions could be established as part of larger surveillance for severe bacterial infections including meningitis and sepsis. Surveillance for invasive bacterial diseases can be improved by: a) applying standardized case definitions; b) create, disseminate and implement standard pediatric guidelines for treatment of pneumonia, empyema and other syndromes associated with invasive bacterial diseases; c) requiring report of clinical laboratory specimens from normally sterile sites that are culture-positive for Hib, S. pneumoniae, N. meningitidis and other invasive bacterial pathogens; and d) implementing standard operating procedures that maximize capacity for detection of invasive bacterial pathogens in hospital laboratories. Future prospective studies of empyema will benefit from standardized case definitions and coding practices for empyema as well as pleural effusion.

The authors declare that they have no competing interests.

BN and PEK conceived and designed the study, assisted with data collection, performed the data analyses and drafted the study manuscript. DEY, DDA, C–HC and XS implemented standardized methods for hospital data collection, verified data sources and accuracy and participated in writing of the study manuscript. LJ, TLN, HLB and WPH provided input into data collection, reviewed outputs from data analysis and assisted in editing of the study manuscript.