Recurrent wheezing in neonatal pneumonia is associated with combined infection with Respiratory Syncytial Virus and Staphylococcus aureus or Klebsiella pneumoniae

Both viral and bacterial infections can be associated with wheezing episodes in children; however, information regarding combined infections with both viral and bacterial pathogens in full term neonates is limited. We sought to investigate the effects of viral–bacterial codetection on pneumonia severity and recurrent wheezing. A retrospective cohort study was conducted on neonates admitted to our hospital with pneumonia from 2009 to 2015. Of 606 total cases, 341 were diagnosed with RSV only, and 265 were diagnosed with both RSV and a potential bacterial pathogen. The leading four species of bacteria codetected with RSV were Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus and Enterobacter cloacae. Neonates with RSV and a potential bacterial pathogen were significantly more likely to have worse symptoms, higher C-reactive protein values and more abnormal chest x-ray manifestations with Bonferroni correction for multiple comparisons (P < 0.01). On Cox regression analysis, an increased risk of recurrent wheezing was found for neonates positive for RSV–Staphylococcus aureus and RSV–Klebsiella pneumoniae. Our findings indicate that the combination of bacteria and RSV in the neonatal airway is associated with more serious clinical characteristics. The presence of RSV and Staphylococcus aureus or Klebsiella pneumoniae may provide predictive markers for wheeze.

Data collection. Data on patient demographics, microbiology results, laboratory parameters and chest x-ray findings were obtained from the hospital electronic medical records system. We used questionnaires to obtain information on the subsequent development of recurrent wheezing, rhinitis and eczema during the first 3 years of life. The questionnaires were based on the Japanese version of the International Study of Asthma and Allergies in Childhood (ISAAC) Phase Three Questionnaire 17 . During the follow-up period, we called parents and asked them questions according to the questionnaire. Wheeze was defined as wheezing or whistling sounds, persistent troublesome cough or breathlessness 6 . A wheezing episode was defined as a respiratory episode with wheezing for more than 1 day. The interval between two episodes was defined as a period of at least 7 days without respiratory symptoms. Recurrent wheezing was defined as having three or more episodes of wheezing 18 . Survival time of wheeze for each child was defined as time from neonatal pneumonia to the first occurrence of wheeze.
Statistical analysis. Continuous variables were analysed using the Mann-Whitney U test because data were not normally distributed. Categorical variables were analysed using the chi-square test, and multiple comparisons were adjusted using the Bonferroni correction (P < 0.01). The association between codetection of virus with bacteria, covariates, and subsequent development of recurrent wheezing, rhinitis and eczema were assessed by logistic regression analysis to estimate odds ratios (ORs). Considering the inconsistent follow-up time, survival statistics were used to analyse the risk of recurrent wheezing. The cumulative risks of recurrent wheezing during the first 3 years stratified according to codetection by RSV and bacteria were estimated by the Kaplan-Meier estimator. Changes in risk were quantified by Cox regression.
Continuous variables were reported as medians (25th-75th percentiles), and categorical variables as frequencies (percentages). Statistical analyses were performed using the IBM Statistical Package for Social Sciences (SPSS) version 21. P < 0.05 was considered statistically significant.

Data availability statement.
Requests for materials should be addressed to Q.L (email: qilu_qi@163.com)

Results
During the study period, a total of 27,169 neonates were admitted to the hospital for clinical pneumonia. Of these, 8128 (29.9%) had nasopharyngeal aspirates collected to check for respiratory viruses within 24 hours of admission.
Other viruses were identified in 50 cases, including 33 samples with parainfluenza virus, 13 with influenza virus and 4 with adenovirus. Escherichia coli was codetected with these viruses more often than other bacteria, with a total of 11 cases. 606 neonates were followed up, and 554 (91.4%) of them had follow-up results during the first 3 years of life.
Clinical Characteristics. In consideration of sample size, RSV-positive cases were further analysed. We assessed whether the main four species of codetected bacteria in RSV-positive neonates influenced disease severity. There were no significant differences in the baseline characteristics of age, sex, weight, gestational age at birth, birth weight, and caesarean section rate between RSV-positive neonates with and without bacterial pathogen codetection with Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, or Enterobacter cloacae. Furthermore, the rates of nasal obstruction, cough, cyanosis, moist rales and diarrhoea did not differ between the groups. Neonates with codetection of RSV and a potential bacterial pathogen were significantly more likely to present with shortness of breath (P < 0.001), wheezing (P = 0.008), chest retraction (P = 0.009) and a higher oxygen requirement (P < 0.001), higher C-reactive protein values (P < 0.001) and more abnormal chest x-rays (P < 0.001) than those with detection of RSV only.
Neonates with codetection of RSV and Staphylococcus aureus had higher white blood cell counts, higher blood platelet counts, and a higher incidence of fever. However, clinical symptoms and laboratory tests were not significantly different between the RSV only and the RSV-Enterobacter cloacae group. After adjustment for multiple comparisons by Bonferroni correction (P < 0.01), neonates with RSV and a potential bacterial pathogen were significantly more likely to have worse symptoms, higher C-reactive protein values and more abnormal chest x-ray manifestations (Table 1).
Codetection in relation to recurrent wheezing, rhinitis and eczema. After controlling for confounding variables with logistic regression analysis, neonates with codetection of RSV-Staphylococcus aureus and RSV-Klebsiella pneumoniae had increased risks of recurrent wheezing (OR 7.16, 95% CI, 2.70-18.94, P < 0.001 and OR 3.66; 95% CI, 1.49-9.94, P = 0.005; respectively) compared with those with RSV only. There was no statistically significant difference among the groups in the risks of rhinitis and eczema (Fig. 2).
Survival analyses for risk factors associated with recurrent wheezing during the first 3 years of life. The risk of recurrent wheezing was increased in neonates with codetection of RSV-Staphylococcus aureus and RSV-Klebsiella pneumoniae, but not in those with codetection of RSV-Escherichia coli or RSV-Enterobacter cloacae ( Table 2). Those with codetection of RSV-Staphylococcus aureus were 4.22 times more likely to have recurrent wheezing than RSV only cases (HR, 4.22; 95% CI, 2.02-8.83). The hazard ratio for the codetection of RSV-Klebsiella pneumoniae was 3.15 (95% CI, 1.51-6.57) for recurrent wheezing.
In the Kaplan-Meier analysis, the risk of recurrent wheezing associated with codetection of RSV-Staphylococcus and RSV-Klebsiella pneumoniae continually increased during the first 3 years of life (Fig. 3). There

Discussion
RSV was the predominant viral pathogen seen in our study, consistent with results from other studies 19,20 . RSV was detected in 10% of neonatal pneumonia samples, however, other studies have shown RSV to be present in about 22%-29% of all acute lower respiratory tract infection cases 19,20 . This could be dueto the fact that we only included full term infants in our study. Infants are most susceptible to RSV infections at 6-11 months of age, when they lack protection from maternal immunoglobulins 19 .  22 . Our findings differed from those studies, showing RSV was more likely to be codetected with Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus and Enterobacter cloacae. This may be due to differences in geography and population, as the most common gram-negative organisms in China are Escherichia coli, Klebsiella pneumoniae and Haemophilus influenzae, and the most common gram-positive organism is Staphylococcus aureus 14 .
In our study, codetection of RSV and potential bacterial pathogens contributed to enhanced symptoms of disease, higher C-reactive protein values and more abnormal chest x-ray manifestations. These findings suggest that during RSV infection, these specific bacterial pathogens were cofactors that contributed to the severity of respiratory symptoms and inflammatory reactions. One possible explanation for this pattern was that virally induced inflammation might alter bacterial gene expression, leading to a more pathogenic phenotype, increased bacterial virulence and decreased clearance of bacteria 23,24 .
The incidence of wheezing is quite high in the first 3 years of life 25 , and RSV might be the initiator (post-RSV wheezing disorder). Notably, there is increasing evidence that the combination of specific bacteria and rhinovirus is associated with wheezing and asthma. Previous studies showed that the detection of rhinovirus together with M. catarrhalis and S. pneumoniae increased the risk of asthma exacerbations in children 4-12 years of age 26 . In our study, cases with codetection of Staphylococcus aureus and Klebsiella pneumoniae with RSV were associated with an increased risk of recurrent wheezing compared with RSV only cases in children from birth to age 3. This may result from the fact that specific bacteria are involved in allergic inflammation and may induce immunomodulatory effects and a Th2-type of eosinophilic inflammation in patients with allergic disease 12,27 .
To our knowledge, this study is the first to research both viral and bacterial detection in full term neonates, and we found a relationship between RSV and specific bacteria. However, some potential study limitations need to be discussed. First, single-centre retrospective cohort analysis has known inherent limitations, and further prospective multi-centre studies are recommended. Second, we had no information on viral and bacterial loads, and quantitative analysis could not be performed in our study. Third, sicker neonates were more likely to have hospital-associated acquisition of bacteria, and, therefore, the bacteria could simply be a marker of disease severity and sequelae. Furthermore, it is worth noting that detection of bacteria in sputum samples might not indicate causative agents. Bacteria may be contaminants from the nasopharynx, because healthy infants often carry pathogenic bacteria there. Thus, because we must be cautious in using the terms 'causative agent' or definite 'coinfection' , we used 'codetected' to state the relationship between the detected bacteria and viruses in our study.

Conclusion and Prospection
The combination of bacteria and RSV in the neonatal airway was found to be associated with an additive effect that resulted in more serious clinical symptoms. The presence of RSV and Staphylococcus aureus or Klebsiella pneumoniae in the airway were associated with an increased risk of recurrent wheezing in early life and may provide a predictive marker for wheezing.