Respiratory viruses from hospitalized children with severe pneumonia in the Philippines

Infant and under-five mortality rates are commonly considered as important health indicators. In 2000, the United Nations introduced the Millennium Development Goal 4 (MDG4) to reduce under-five mortality by two-thirds between 1990 and 2015. The under-five mortality rate has steadily decreased from 90 per 1000 live births in 1990 to 65 per 1000 live births in 2008 [1]. However, the rate remains high in many developing countries[1], and pneumonia is still the leading cause of childhood mortality [2, 3],particularly in the developing countries. It is estimated that approximately 2 million children die from pneumonia annually, and 90% of these deaths occur in developing countries [2‐4]. However, viral etiologies may be underestimated in cases of severe pneumonia in the resource-limited countries with limited diagnostic facilities. The role of viral pathogens in severe pneumonia has come into prominence as the role of bacterial infection decreases through early case detection, appropriate antibiotic treatment [5] and introduction of conjugate vaccines.

Historically, the common viruses most closely associated with childhood pneumonia have included influenza virus (Flu), respiratory syncytial virus (RSV), parainfluenza virus (PIV), and adenovirus (AdV). The development of molecular techniques facilitated the detection of novel viruses in patients with respiratory infections. Human metapneumovirus (hMPV) was first documented in 2001 [6]. Since then, a series of human coronaviruses (HCoVs), including NL-63 [7], -NH [8], and -HKU-1 [9], a new genogroup of human rhinoviruses (HRVs) called type-C rhinoviruses [10], human bocavirus (HBoV) [11], and human polyomaviruses WU (WUV) and KI (KIV) [12, 13] have been reported. We recently demonstrated the presence of hMPV, HCoV-NL63, HCoV-HKU1, HBoV, WUV, and KIV in patients with influenza-like illness in the Philippines [14], but were unable to assess their clinical significance in children with pneumonia. The etiological roles of these novel viruses remain undefined.

In this prospective study, we investigated the association between presence of a virus and severity of disease in hospitalized children with severe pneumonia to provide an indication of causality. With this aim, we detected viral as well as bacterial pathogens in children hospitalized with severe community-acquired pneumonia and assessed the clinical significance of these pathogens.

Our study highlighted the potential importance of viruses as pathogens of community-acquired pneumonia among hospitalized children in the Philippines. The CFR in our study setting was as high as 8%, which indicated that childhood pneumonia is still an important cause of death in the Philippines. We conducted our current study following the IMCI guideline, because IMCI has been implemented in various developing countries for nearly 20 years [31] as well as we also investigated the causative pathogens in those hospitalized children. Because, the IMCI approach toward childhood pneumonia focuses on bacterial infection, mainly through the empirical use of antibacterial drugs; however, the approach to the management of viral infections is not included in IMCI [19]. Furthermore, IMCI was developed on the basis of etiological studies conducted before 1990, when the detection methods for viral pathogens were suboptimal [32‐34]. However, modern molecular methods, such as PCR exhibits higher sensitivity in virus detection. In addition, these methods facilitate the discovery of several new viruses as possible etiological agents for respiratory infection [6‐13]. Using these methods, at least one virus was detected in 60.8% of cases in this study. Recent etiological studies also detected viruses in more than 50% of cases in similar settings [35, 36].

Although we have demonstrated an important potential role of viral pathogens, economical and effective options for the prevention and treatment of viral infections that are practical to implement in developing countries are unavailable. HRV- and RSV-positive cases accounted for approximately 55% of our study population and had significant effects including those on mortality. However, no effective and safe vaccines or antiviral drugs for these viruses are currently available [37]. The monoclonal RSV antibody for prophylaxis has been used for high-risk infants in developed countries, but this antibody is far too expensive to be used in developing countries [38]. FluA had the highest CFR and was a significant risk factor for fatal outcome, which suggests its potential high mortality impact during larger epidemics. Antiviral drugs such as neuraminidase inhibitors and vaccines for influenza are widely available and used in developed countries; however, they are not affordable in most of the developing countries [39]. To reduce childhood mortality due to viral pneumonia, feasible and inexpensive management strategies, such as triaging hypoxic patients with using a pulse oximeter and administering early treatment with oxygen, should be urgently developed [40].

Although we previously demonstrated the presence of novel viruses in the Philippines [14], our current study provided additional insight on novel viruses. One of the significant findings was the detection of HRV-C. In previous studies, the impact of HRV-C on morbidity and mortality was similar to that of HRV-A and was much greater than that of HRV-B [41‐43]. In the present study, HRV-C, the prevalence of which was less than that of HRV-A and greater than that of HRV-B, was detected in 10.1% of the total cases and in three fatal cases. These results suggest that HRV-C could be an important viral pathogen in the cause of severe childhood pneumonia in the Philippines. However, its etiological role is still controversial because this virus has also been detected in healthy individuals [42]. Further studies are necessary to define the role of HRV-C in childhood pneumonia. Another novel virus that shares similar epidemiological and clinical features with RSV in industrialized countries is hMPV [44]. Previous studies in other developing countries indicated that the incidence of hMPV was lower than that of RSV [35, 36, 45]. However, we observed a fatal case of hMPV which might have an impact in our study setting. A series of HCoVs have been discovered by recent techniques [7‐9]. Although none of the HCoV-positive cases exhibited improvement during hospitalization, the number of positive cases was very small. Conversely, there was no fatal case among those positive for newly discovered DNA viruses including HBoV, WUV, and KIV which suggest that their impact in our study is insignificant. However, their severity should be carefully interpreted because of the small number of positive cases for these viruses.

Studies on childhood pneumonia were conducted in the Philippines in the 1980s and 1990s [32, 33, 46]. In addition to those previous studies, subtyping of viral genetic information revealed group-specific epidemiology. Tupasi et al. demonstrated that RSV was the most common viral cause among children admitted with acute lower respiratory infection in the Philippines [33]. This was consistent with our present study as well as with other recent studies in developing countries [35, 36, 45, 47]. RSV can be classified into two major groups; however, their proportions may change during epidemics [48]. Similarly, we demonstrated that the increase in the number of cases between September and December 2008 was primarily due to RSV-A, although RSV-B was also present in the early stage of the epidemic. We may further need to assess the RSV itself to understand the circulation dynamics of the virus [49]. On the contrary, Capeding et al. demonstrated that AdVs were also important viral pathogens in the Philippines [46]. We observed clustering of AdV-7 in genus B and AdV-6 in genus C, both of which are known to cause respiratory infections of varying severities among children [50]. As each genus of AdVs has different clinical and epidemiological features [51], genetic classification will help us understand the epidemiology of AdVs in tropical climates.

Bacteremia was observed in 3.8% of our cases, and this rate is similar to our previous findings in the Philippines (2.9% and 5%) [20, 22]. In addition to two major bacterial pathogens, S. pneumoniae (n = 4) and H. influenzae (n = 1), the most commonly isolated bacteria was B. cepacia group (n = 9). B. cepacia is a habitant of the natural environment, but can cause severe infection in the hospital settings especially among immunocompromised patients in ICU [52, 53]. Since B. cepacia infection in the community is rare [54, 55] and the samples were collected upon admission, it is more likely that those isolates of B. cepacia group are the contaminants due to contaminated sampling devices, such as disinfectant products [56], rather than causative bacteria pathogen for the pneumonia. However, it should be noted that 3 of those B. cepacia group positive-cases died. However, further investigation on the pathogenicity of the B. cepacia in our study setting is required.

We might have underestimated the incidence of bacterial pneumonia since prior antibiotic use was observed among 32.4% [264/814] of the enrolled cases. However, a recent study indicated that the prevalence of bacteremia among children with community-acquired pneumonia in the emergency department was as low as 2.1% [57]. Unlike viral isolation, there is no “gold standard” for the diagnosis of bacterial pneumonia among children [58]. Blood culture is widely performed, but its sensitivity is as low as 1-3% [59]. Bacteria cultured from the nasopharynx or throat do not always represent the pathogens in the lungs, as it may include the normal flora of the nose or throat. Lung tap provides an accurate diagnosis of childhood pneumonia [60], but it is an invasive procedure, difficult to perform aseptically, and therefore it is not commonly practiced in developing countries to avoid any iatrogenic complication due to the procedure itself.

The treatment approach of IMCI recommends the empirical use of antibacterial drugs [19], but there is a dilemma of the overuse of antibiotics that increases the emergence of drug resistance. Moreover, antibiotics can be purchased without prescription in the Philippines. For S. pneumoniae, all three invasive isolates were susceptible to penicillin, as reported in a previous study in Bohol, Philippines [61]. However, two cases with MRSA did not exhibited improvement during hospitalization. Community-acquired MRSA is a growing problem in developing countries [62]. As IMCI encourages vaccine and antibiotic use for bacterial infections, continuous monitoring of serotype distribution and drug susceptibility is essential for devising a local and cost-effective strategy against childhood pneumonia.

There are some limitations of our study. Firstly, the detection of viruses by molecular techniques may not necessarily indicate that the detected virus is a causative agent of respiratory symptoms since it does not fulfill Koch’s postulate. We understand that the gold standard for the diagnosis of viral infection is virus isolation; however, many novel viruses cannot be isolated, and demonstrating the presence of the viral genome is the only available detection method. On the other hand, we identified multiple viruses in different combinations especially among patients positive for HBoV, WUV, and KIV (Table 2). WUV and KIV are members of Polyomaviridae that frequently cause latent infection along with reactivation by immunosupression [63]. Their pathogenicity needs to be elucidated with further epidemiological and clinical studies including appropriate “healthy” control groups. Secondly, we were not able to obtain the demographic information of those who refused to participate in the study due to ethical concerns. The proportion of very severe pneumonia among those who refused to participate was smaller than that of admitted cases, which may have led to an underestimation of the severity of the disease. Thirdly, our study was conducted in the only referral hospital in the region. Therefore, our study population may represent only severe and very severe cases with access to healthcare facilities in the study area. This may have resulted in the underestimation of some etiologies. FluA had the highest CFR, but its prevalence was low. This may be due to the reason that the time interval between onset and admission was 4 days, whereas the viral titer of FluA in respiratory specimens was highest within 48 hours of onset [64].

Our study demonstrates a potentially important role for viral pathogens in the etiology of severe pneumonia among the children in the Philippines. Our findings suggest that viral etiologies should be always kept in mind while developing effective strategies to reduce child pneumonia-related deaths in developing countries.