Pneumonia remains a leading cause of mortality globally: 4.2 million pneumonia deaths were recorded in 2004 [1]. The highest incidence of disease occurs in young children [2,3]. An estimated 156 million pneumonia episodes occur annually in children less than five years old. Over 95% of these occur in the developing world, where the incidence of clinical pneumonia is estimated to be 0.29 episodes per child-year. Almost three-quarters of childhood pneumonia deaths occur in sub-Saharan Africa and Southeast Asia [4]. Bacterial pathogens, most notably Streptococcus pneumoniae and Haemophilus influenzae type B, are important vaccine-preventable causes of pneumonia [5]. Viruses, in particular influenza and respiratory syncytial virus (RSV), are also responsible for a large number of pneumonia cases each year [6]. Using global population data for 2005, for children under the age of five years, it was estimated that RSV was responsible for over 30 million episodes of lower respiratory tract infections (LRTI), with ~3 million of these requiring hospital admission, and 66,000-199,000 deaths [7]. By similar analyses of data from 2008, influenza viruses were estimated to cause 20 million LRTI and 1 million severe LRTI, with 28,000-111,500 deaths, in children aged less than five years [8]. In both of these reviews, 99% of deaths from either influenza–or RSV–associated LRTI occurred in the developing world. These viruses may be responsible for up to 35% of LRTI (RSV 22% [7] and influenza 13% [8]) in children under the age of five years. Other viral pathogens associated with childhood pneumonia include adenoviruses, human metapneumovirus (hMPV), and parainfluenza viruses [5,9].

Approximately one-third of the worldwide refugee population of 15 million live in camps [10]. These camps are often crowded with poor sanitation, providing ideal conditions for transmission of respiratory pathogens [11,12]. There have been refugees from Myanmar (Burma) living in camps in Thailand since 1984. In 2006, lower respiratory tract infections were estimated to be the cause of 9% of deaths, and were responsible for 25% of all reported morbidity, in the under-5 age group of the border refugee population. The overall under-5 year mortality rate was 6 per 1,000, giving an estimated LRTI-specific mortality rate of 0.5 per 1,000 [13].

In 2007, the US Centers for Disease Control and Prevention (CDC) and the Shoklo Malaria Research Unit (SMRU) established a respiratory virus surveillance programme in the Burmese refugee population living in Maela camp, Northwest Thailand. The programme included patients admitted to hospital with pneumonia during April 2009-September 2011. The aim of in-patient surveillance was to determine the relative burden of virus-associated pneumonia. The results of 30 months of in-patient surveillance are presented here.

Among all IPD pneumonia admissions, 835 patient episodes were enrolled in enhanced surveillance. After review of the symptom questionnaire, 117 patient episodes were excluded because of failure to meet the case definition and one episode was excluded for a patient who presented twice within 14 days. In three episodes, NPA specimens were not collected and in another six, the specimens were technically inadequate (internal control PCR negative). The remaining 708 patient episodes were included in the following analyses.

Laboratory-enhanced surveillance has documented the contribution of respiratory viruses to 708 hospitalised clinical pneumonia episodes occurring in a crowded refugee camp on the Thailand-Myanmar border during April 2009 to September 2011. As expected, the vast majority of patients were aged less than five years [2]. Viruses were detected in the nasopharynx of half of the cases. Virus detection was significantly more likely in infants compared to older children and adults, although the number of adult cases was small. During the same period, routine hospital surveillance detected 2,081 in-patient LRTI episodes, including 32 (1.5%) deaths and 1,340 (64.4%) episodes in children under five years [21]. Therefore, IPD pneumonia episodes under enhanced surveillance reported here accounted for 34.0% of all IPD LRTI episodes (47.8% of IPD LRTI episodes in children aged less than five years and 9.2% of episodes in persons aged five years or older).

The Maela data add to the scant data on the aetiology of pneumonia in refugee populations. The results are broadly consistent with a similar surveillance programme conducted in two Kenyan refugee camps [22], where 51.3% patients with severe acute respiratory infection (SARI) had at least one of adenovirus, hMPV, influenza A/B, parainfluenza virus 1-3, or RSV detected. In the Kenya surveillance, adenovirus was the commonest virus detected (22.7% of specimens), followed by RSV (14.1%). An influenza virus was detected in 11.6% of specimens. The majority (82%) of hospitalised SARI cases were children less than five years of age.

Results from pneumonia aetiology studies from various locations in the developing world have confirmed the high prevalence of virus detection in hospitalised pneumonia episodes in young children [23,24]. RSV has been frequently identified as the commonest virus and is associated with severe infections [25-28]. A significant proportion of RSV infections occur in the first year of life, although both primary infection and re-infections are common in older children [26,29,30]. RSV has previously been documented to be a significant pathogen in Maela, with an incidence of RSV-associated pneumonia of 0.24 episodes per child-year at risk in a cohort of infants followed from birth until two years of age [31,32]. Our results confirm that RSV is the leading virus associated with pneumonia in the general Maela camp population. The majority of infections were in young children and only two (2/176, 1.1%) RSV infections occurred in patients aged five years or older.

Influenza viruses were detected in almost 10% of patient episodes of pneumonia. This figure is consistent with previously published data on influenza hospitalisations. Simmerman and Uyeki determined that influenza viruses were detected in 6-14% of hospitalised pneumonia cases in a recent review of East and Southeast Asian data [33].

The role of adenoviruses in the aetiology of pneumonia remains unclear. Frequent re-infection and persistence in young children makes their detection in NPA specimens at the time of pneumonia diagnosis difficult to interpret [34,35]. A recent case–control study from Kenya, undertaken as part of a multi-centre paediatric pneumonia aetiology study (PERCH), did not find a higher odds of adenovirus detection in pneumonia cases compared with controls [36]. In that study, only the detection of RSV in the upper respiratory tract had a significant association with hospitalisation for pneumonia.

WHO definitions for clinical pneumonia in childhood were used in our enhanced surveillance in Maela. These definitions were designed to have optimal sensitivity for the diagnosis of potentially life-threatening bacterial infection in resource-poor settings [15]. It was accepted that some over-diagnosis and unnecessary treatment would occur [37]. As was demonstrated in a South African study, where virus-associated pneumonias were reduced by a third in children given a nine-valent pneumococcal conjugate vaccine, a proportion of virus-positive individuals would have had a secondary bacterial infection [38], but many will have received an unnecessary course of antibiotics. Recent work from Pakistan has confirmed that placebo and amoxicillin had equivalent efficacy for WHO non-severe pneumonia and that oral amoxicillin at home is an acceptable alternative to hospital admission and parenteral antibiotics for severe pneumonia [39,40]. In Maela, almost all patients were treated with at least one antimicrobial drug, yet over a third (excluding adenovirus) had a proven viral infection. Collectively, these findings point to the likely over treatment of viral infections in children with WHO defined pneumonia, which may contribute to the rising prevalence of antimicrobial resistance in the developing world [41].

The enhanced surveillance system had several limitations. Not all hospitalised pneumonia episodes were captured and therefore it was not possible to calculate virus-specific incidence rates. Estimating the representativeness of the patients enrolled by comparison of enhanced surveillance results with routine surveillance figures is problematic. However, as we note, if a direct comparison is made, enhanced surveillance included 47.8% (640/1,340) of all LRTI episodes identified through routine surveillance in children under five years of age but only 9.2% (68/741) of episodes in patients aged five years or older. Differences in case definitions used in the routine surveillance compared with enhanced surveillance is the most likely explanation for this discrepancy [21]. For the under-5 age group the definitions were almost identical (i.e. WHO-based) but for the older age group LRTI was defined in routine surveillance by the presence of a fever plus cough or sore throat and shortness of breath/difficulty breathing. Absence of a requirement for abnormal chest signs likely resulted in inclusion of many non-pneumonia cases within the LRTI category. The low number of deaths recorded in the HIS system suggests that severity was not a significant reason for non-inclusion. Also, since PU-AMI hospital was the only general medical admissions unit within the camp, there were not significant numbers of patients missed as a result of admission to other hospitals. However, it remains possible that the older participants were not representative of the entire hospitalised pneumonia patient group. Additional severity data (e.g. need for supplemental oxygen and length of stay) and outcome data were not collected, which further limit the conclusions that can be drawn from the surveillance and the comparisons that can be made with other studies. The panel of viruses tested for included the key pathogens for which evidence of an association with pneumonia is proven, but inclusion of PCR assays to detect additional respiratory viruses would have added value and may have identified a higher prevalence of multiple infections. Studies using multiplexed virus PCR assays have detected both an increased proportion of children with single and multiple viral lower respiratory infections [42]. Both human bocavirus and rhinoviruses may be detected in a large proportion of pneumonia cases, although data regarding causality from case–control studies are still limited [36,43]. Despite this, it was demonstrated that influenza virus and/or RSV were associated with a third of hospitalised pneumonia episodes in Maela.

The cost of in-patient pneumonia treatment is high. In two recent studies, the estimated average cost per district hospital pneumonia admission was US$99.26 in Kenya and US$490.80 in Thailand [44,45]. Influenza infections are vaccine preventable. A vaccine to prevent RSV infection remains elusive, although progress continues to be made [46,47]. Inclusion of influenza immunisation, and RSV immunisation should it become available, in the immunisation programme for refugees on the Thailand-Myanmar border would likely significantly reduce the burden of pneumonia requiring hospitalisation.

Viruses were commonly identified in Burmese refugees admitted to hospital with clinically-diagnosed pneumonia. Use of influenza immunisation and the development of vaccines targeting other common respiratory viruses would be likely to reduce the incidence of pneumonia in children living in refugee camps and similar settings.

The authors declare that they have no competing interests.

PT, VC, CT, CD, CP, LO, and FN conceived the surveillance project. NC and CT were responsible for specimen and data collection. WW performed the laboratory work. PT and VC did the data analysis. PT prepared the first draft of the manuscript. All authors reviewed and contributed to revisions of the manuscript. All authors read and approved the final manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-2334/13/434/prepub