Severe lower respiratory tract infection in infants and toddlers from a non-affluent population: viral etiology and co-detection as risk factors

Lower respiratory tract infections (LRTI) represent an important public health burden in the first years of life accounting for approximately one fifth of all deaths in children below five years of age, especially in developing countries [1]. The specific role of newly identified viruses on LRTIs, like Human Metapneumovirus (hMPV), has been studied in recent years [2]. However, its impact among non-affluent populations has been scarcely evaluated. In such locales, infants with respiratory syncytial virus (RSV)-associated LRTIs present a three times greater risk of a fatal event, when compared to their peers in developed countries [3].

Although RSV is well recognized as the main agent associated with severe LRTIs, recent data indicate that other viruses may play a significant role in these clinical outcomes. Human rhinovirus (HRV) seems to be of particular interest, as the most prevalent virus in respiratory illnesses even in the first years of life [4, 5], being associated with severe acute bronchiolitis, especially among children of atopic parents [6]. Moreover, a recent study showed that, in a population of preterm infants, HRV was the most prevalent agent associated with severe bronchiolitis [7]. Also of interest is the fact that wheeze-related HRV infection in the first year of life is associated with an increased risk for developing asthma later in life [8], and that this effect was greater than the observed in relation to RSV [9].

The impact on severity of early life respiratory infections may be also affected by viral co-detections diagnosed through sensitive PCR analyses. Some studies have shown a positive association between viral co-detection and worse clinical outcomes [10, 11], while others have failed to show results in the same direction [12‐14].

The aims of our study were to determine the current impact of newly identified viruses on the severity of LRTI in infants seen in the emergency room and pediatric wards from a tertiary hospital in a developing country, and how specific viruses alone or in co-detections increased the degree of clinical severity of disease.

Two hundred and sixty patients were enrolled in the study. Characteristics of the patients recruited are presented in Table 1. In the whole sample there were more boys than girls, almost half of children were younger than six months, 24% of all children were born premature, 29% had been exposed to tobacco during gestation, and more than a half had a family history of asthma (parents or siblings). Overall, median hospital length was 6 days and median supplemental oxygen requirement was 5 days.

The presence of viruses was detected by PCR in 222 (85%) of all LRTI episodes. Co-detection was present in 146/260 (56%). The frequencies of viral detection and co-detection are shown in Table 2. The most common single infection was related to RSV (54%), followed by hMPV (32%), and HRV (21%).

During the surveyed time RSV has shown an incidence peak in the beginning of the cold season (i.e. from april to November) in the southern hemisphere, followed in late winter by peaks of hMPV, HRV, and HBov (Figure 1). Other viruses such as HPIV and Human Adenovirus (HAdv) showed low but constant rates throughout the season.

In the univariate analyses, length of hospital stay and need of supplemental oxygen were significantly associated with age (≤6 months), maternal smoking during pregnancy and with family history of asthma (parents and/or siblings) (Table 3). Infants younger than 6 months of age stayed in hospital 3.8 days longer than older infants (p < 0.001), and those with a family history of asthma stayed 2.4 days longer than those without a family history of asthma (p < 001). Also, infants 6 months of age or younger needed supplemental oxygen for an extra 3.8 days, when compared to older infants/children. A similar finding was observed for children with a family history of asthma, who required 2.4 extra days of oxygen compared to those without the family history of asthma (Table 3). Other risk factors, such as breastfeeding, indoor smoking, current parental smoking, siblings, and overcrowding were not significantly associated with neither of the main outcomes.

Infants with positive PCR for HRV alone as well as those co-detected with RSV and HRV also had significantly longer hospital stays (3.2 days, p = 0.001; and 5.5 days, p = 0.002, respectively) than those with other detected viruses. Extended time in use of supplemental oxygen was also associated with HRV (2.8 days, p = 0.002) and RSV (3.7 days, p = 0.013), but also with Influenza virus A or B (2.2 days, p = 0.042), when compared to those with other viruses in single or in co-detection.

Infants with HRV-LRTIs stayed an extra 2.2 days in hospital (p = 0.011), for a total of 7.7 (95% CI: 6.1-9.3) days when compared to those with other infections, after adjusting for potential confounding variables (Table 4). Table 5 shows that infants with combined HRV and RSV positive PCR in the same samples stayed 4.5 extra days (p = 0.004) than those without HRV and RSV in these adjusted models (that included sex, age ≤ 6 versus >6 months, prematurity, family history of asthma and maternal smoking during pregnancy), (95% CI: 7.0-13.0) days. Time in use of supplemental oxygen followed the same association trends. Infants with proven RSV infections needed 4.75 (95% CI: 3.97-5.53) extra days of oxygen, while those with HRV used supplemental oxygen for 1.4 extra days, and those with RSV and HRV co-detection for 2.2 days, when compared to infants presenting positive PCRs for other viruses alone or in combinations. Figure 2 illustrates the association between HRV and RSV + HRV co-detection with increased length of hospital stay and oxygen use, and this effect is especially significant for infants younger than 6 months of age. Influenza viruses were not associated with longer use of supplemental oxygen when controlling for the demographic variables.

Our results suggest that infants with severe LRTI and positive PCR for HRV, alone or in co-detection with RSV, stayed hospitalized longer periods and utilized more supplemental oxygen, when compared to children infected by other viruses, including those with RSV-alone. Our data also reinforce previous findings that identified RSV as the major agent associated with severe LRTIs among children in a hospital setting, in populations of low socio-economic status, where other environmental and social variables potentially play a role [16, 17]. Infants younger than six months and those with a family history of asthma/recurrent wheeze are also at greater risk for disease severity.

RSV was the most frequently detected virus, accounting for a high burden of LRTIs in our population. Although it is not possible to establish an unequivocal correlation between LRTIs and upper airway viral detection, the finding of RSV in over 50% of hospitalized children in our study strongly suggests that its impact is still indeed very high in this region, regardless of the presence of newly identified viruses. These results are in accordance with recently published studies in Brazil, which also identified RSV as the main agent responsible for severe LRTI, especially in a hospital setting. Nascimento and coworkers have shown an overall viral detection rate of 93% from nasopharyngeal samples in a small group of children below 2 years of age, and reported RSV as the most prevalent virus (63.6%) [19]. Another study aiming to investigate the role of HBov and hMPV in LRTIs in southern Brazil also showed similar results (at least one positive virus in 90% of the samples and RSV positive in 49.3%) [20]. The high detection rate of RSV in children with LRTI in a hospital setting such as ours is consistent with most studies worldwide and the burden due to viral respiratory disease seems as high in these locales as they are in more developed countries. A recent analysis of children admitted into the hospital due to acute bronchiolitis in Texas, USA, has shown a steady increase in admissions over a 5-year period and this has been credited mostly to RSV [21]. This increasing role for severe viral LRTIs, observed also in other studies, is probably explained by a series of complex environmental and social changes that seem to affect how viruses spread in communities.

Some studies have reported different clinical outcomes for specific viruses causing LRTIs, especially in the presence of co-detections, such as with RSV and HBov [22, 23], while others did not reproduce such findings [17, 24, 25] and these associations remain unclear. A lack of association between an overall finding of any viral co-detection and LRTI severity was reported in studies performed in non-affluent countries [2, 12, 26], as well as in developed countries [13, 14]. In our study, patients detected with HRV alone and RSV + HRV presented increased length of hospitalization and increased time of supplemental oxygen use. Papadopoulos et al. has shown a five-fold increase in clinical severity in infants with acute bronchiolitis due to HRV, compared to those infected only with RSV. Compared to those with positive RSV samples without HRV co-detection, infants with HRV were older, had lower birth weights and were hospitalized earlier [16]. Some explanations for apparently contradictory findings in a myriad of studies could be attributed to the lack of uniform criteria for subject inclusion and standardized statistical analysis [14]. Other plausible explanations are the natural variation in HRV prevalence in different seasons and possible variations in the prevalence of type C HRV. This agent is associated with more severe disease and was already described as a major cause of LRTI in infants from non-affluent countries [27, 28]. Unfortunately, in our study, we were not able to determine the prevalence of HRV subtypes, and this stands as an interesting subject for further research.

The association of HRV (alone or with RSV) with LRTI severity and atopy has not been widely studied. The relationship between persistent wheezing at 3 years and at 6 years versus relevant HRV infection in early life is well established [9, 29], but there are few studies looking at these relationships in the first year of life. A recent study suggests that the relationship between HRV in early life LRTI and subsequent recurrent wheeze/asthma is dependent on allergic sensitization, which seems to precede the viral insult in a causal model [30]. This association we have found between HRV with increased severity (using the surrogates of length of hospital stay and days in supplemental oxygen) is a major finding. Hence, the association between HRV infection, increased severity and atopy remains to be better clarified.

It has already been shown that HRV is able to reduce cell proliferation and decreases the self-repair capacity of bronchial epithelial cells [31]. Therefore, our data may suggest that in certain subsets of patients the burden of HRV in acute LRTI should be considered distinct from that of other viruses. Another plausible explanation for our findings could be the possibility of HRV persistence in the airways leading to an “over detection” of the virus, simultaneously with those infected only by RSV [32]. In our study this hypothesis seems improbable since we detected a clear worsening in the clinical markers in patients with RSV that were also detected with HRV.

Another interesting finding of our study was the high prevalence of newly described viruses. hMPV was detected in almost one third of all episodes, but did not seem to affect the main outcomes studied here, in the way RSV and HRV have done. While the recognition of the impact of human hMPV is increasing, its prevalence is still probably underestimated in clinical practice, since laboratory testing has become widely available only in recent years [18, 24, 33]. HBov was also detected in a large number of nasal samples and it was very frequently associated (co-detected) with other agents. In our study, only 2/27 patients with HBov had this virus detected as single agent. The overall HBov detection rate was higher in our data compared to previous studies [34, 35], and this may be explained again by natural seasonal variations. It is also interesting to notice the seasonal pattern of both hMPV and HBov, which present their peak prevalence rates in late winter, right after the RSV peak, which occurs earlier in winter.

The main limitation of our study is the lack of surveillance in consecutive years, which could have biased results in case of an outbreak of one specific virus in a given year. Our eight months of viral surveillance could potentially have failed to detect any atypical outbreak, which did not seem likely to have occurred. Also, reliable tests capable of ruling out bacterial co-detection were not available at the time of the study. This could have underestimated the burden of bacteria in our sample and the co-detection of viruses and bacteria remains an interesting issue for further studies.

In our study, RSV was the most prevalent viral agent in hospitalized patients with LRTI and the co-detection of HRV in patients with RSV infection increased hospital stay and days in use of supplemental oxygen. Interestingly, even in developing countries, the role of recently discovered viruses needs to be further studied in order to identify novel risk factors of susceptibility/severity, and new treatment targets for these agents. We also highlight the role of HRV as an important risk factor for severe LRTI, particularly when simultaneously associated with RSV, which strongly suggests that co-detection may also mean co-infection, since the combination of the two agents seem to affect clinical outcomes. Longitudinal studies with control groups are necessary to confirm these results in populations at greater risk for severe respiratory disease.