Viral etiology of life‐threatening pediatric pneumonia: A matched case‐control study

Abstract Background Pediatric pneumonia remains a significant health challenge, while the viral risk factors for adverse outcomes in pediatric pneumonia are not yet fully clear. Methods A matched case‐control study of pediatric patients with pneumonia was carried out in Beijing, China, between 2007 and 2015. The study enrolled 334 intensive care unit patients who developed life‐threatening diseases and 522 controls matched to the sex, age, ethnicity, admission dates, and residing district of the cases suffered from pneumonia. Nasopharyngeal aspirates were taken from all participants and tested by PCR for 18 common respiratory viruses. Results At least, one virus was detected in 257 (77%) of the cases and 409 (78%) of the controls. We observed no difference in the prevalence of 17 respiratory viruses between cases and controls but found a higher frequency of influenza A virus (IFV‐A) in the cases than in the controls (7% vs 4%, P = .036). After adjusting for comorbid conditions and a history of reactive airway diseases, IFV‐A was associated with an increase in life‐threatening pneumonia (adjusted odds ratio = 2.55, 95% CI = 1.24‐5.24). Young age and congenital heart disease (aOR = 10.16‐10.27, P < .001) were also independent risk factors. Conclusions The prevention of IFV infection is critical in decreasing the risk of life‐threatening pneumonia in children.


| INTRODUC TI ON
Pneumonia is a leading cause of death globally, particularly for children. 1 One in five children worldwide suffers an episode of pneumonia each year, 2 and 12%-25% of pediatric pneumonia patients progress to severe pneumonia requiring hospital referral and treatment. Roughly 21% of children who require hospitalization for pneumonia are admitted into a pediatric intensive care unit (PICU), resulting in death in as many as 1% of cases. 3,4 Due to the health burden of pediatric pneumonia, there is a growing interest in clarifying the etiological agents that contribute to the severity and progression of the disease, and respiratory viruses have been explored in this context.
The respiratory viruses associated with the incidence of pneumonia have been reported. [5][6][7] However, few data described the contributions of viral agents to the development of severe pneumonia among children in inpatient settings. The understanding of the risk factors that predispose pediatric pneumonia patients to life-threatening events may help improve clinical care, guide diagnostic testing, reduce the number of unnecessary antibiotic prescriptions, decrease medical costs, and lower the risk of nosocomial infections in a hospital setting. 8 To clarify the contributions of respiratory viruses to very severe pediatric pneumonia, we conducted a matched case-control study to assess the contributions of several common respiratory viruses.

| Study design
A matched case-control study was designed to analyze a cohort of pediatric patients with pneumonia at Beijing Children's Hospital from April 1, 2007, to December 31, 2015, in Beijing, China. The following three criteria were used to enroll patients: (a) had symptoms of acute infection, including fever (body temperature >38.0°C) or hypothermia (body temperature <35.5°C), leukocytosis (for children <5 years old, white blood cell count [WBC] >15 000/mL, and for children ≥5 years old, WBC > 11 000/mL), or leukopenia (for children <5 years old, WBC < 5000/mL, and for children ≥5 years old, WBC < 4000/mL); (b) had signs and/or symptoms of lower respiratory tract infections, including cough, sputum production, shortness of breath, tachypnea (based on age; <2 months, ≥60 breaths/min; 2-11 months, ≥50 breaths/min; 12-59 months, ≥40 breaths/min; >5 years, ≥30 breaths/min), wheezing or crackles, dyspnea, or chest pain; and (c) had radiograph evidence suggestive of pneumonia or pleural effusion. Patients were excluded if they were younger than 28 days or older than 14 years of age, with known immunosuppressive diseases including solid organ or hematopoietic stem cell transplants, receiving chemotherapy, HIV-positive or had AIDS, and taking steroids for more than 30 days. To avoid duplication, patients who had recurrent hospitalizations within 3 months of enrollment were also excluded.
All hospitalized patients were reviewed retrospectively to determine the assignment of cases and controls. All patients who were admitted into the PICU and developed life-threatening diseases, defined as mechanical ventilation, sepsis, shock, or death, were defined as cases. These patients ranged in ages from 28 days to 13 years.
Control patients were selected from general wards and did not present with life-threatening diseases. For each case, up to two controls were matched according to sex, age (±2 months), ethnicity, admission date (±15 days), and residing district. If no eligible controls were found, the matched conditions were expanded to include patient ±6 months of age and with an admission date of ±4 weeks for patients <5 years of age, or age of ±1 years and an admission date of ±4 weeks in patients more than 5 years of age. Two controls were selected at random if there were more than two eligible controls in the cohort.
To validate the data, two pediatricians and one pediatric radiologist independently reviewed all chest radiographs, important clinical outcomes, and diagnoses. Inconsistent diagnoses were discussed with a third senior pediatrician, who would make a final decision. Clinical outcomes or diagnoses used to define the cases and controls were collected from medical charts at the time of hospital discharge, including PICU admission, non-invasive ventilation, for example, continuous positive airway pressure (CPAP), invasive ventilation including mechanical ventilation involving tracheostomy or endotracheal tube, acute respiratory failure (code 518.81) according to the International Classification of Diseases, ninth revision (ICD-9), shock (ICD-9 code 785.5x), sepsis (ICD-9 code 038.xx), and death. Baseline data at hospital admission, including demographic data (sex and age), epidemiological data (date of illness onset and prematurity history of children defined as gestational age of children <37 weeks), and clinical data (symptoms/ signs and comorbid conditions) were retrieved from study records.
Observed comorbid conditions included congenital heart disease (CHD), defined as children with a diagnosis ICD-9 coded as 745.

| Statistical analysis
We used chi-square tests or Fisher's exact tests to analyze categorical variables and Wilcoxon rank-sum tests to analyze continuous variables as appropriate. A two-sided P-value of < .05 was considered statistically significant. Conditional logistic regression was used to explore the relationship between viral exposure at hospital admission and subsequent life-threatening disease during hospitalization.
The comorbidities, breast-feeding, influenza vaccination last year, and initial antibiotic usage were also included in the multivariable conditional logistic regression to account for potential confounders if they were significant at P < .10 in univariate analysis. Data are reported as crude or adjusted odds ratio (OR) with a 95% confidence interval (95% CI). We also conducted sensitive analysis between different age strata, for example, children aged 1-59 months and children 5-13 years. All analyses were conducted in r version 2.15.3 (R Foundation for Statistical Computing). 10

| Characteristics of subjects
During the study period, a total of 4568 pediatric patients were hospitalized for pneumonia. We identified 347 (8%) PICU admissions with life-threatening events during hospitalization, of which 13 cases were excluded because no eligible matched controls were available, even using the expanding matched criteria (Figure 1)

| D ISCUSS I ON
In contrast, 56%-78% of severe pneumonia is associated with respiratory viruses at the severe pneumonia level, and RSV is considered contributing decreased to 23% (range: 18%-34%) and IFVs to 7%. 1,3,12-16 At the third level, child deaths caused by viral pathogens are low among children with pneumonia, with only 6.6% of deaths being attributed to RSV and IFVs. 17 It is well-known that most mildto-moderate childhood pneumonia is caused by viruses, whereas most severe and fatal pneumonia cases are caused by bacterial infections. 11,17 Although the prevalence of RSV is high in hospitalized children in previous study, 3 deaths associated with RSV are uncommon in high-resource settings, and the majority of these deaths occur in children with chronic conditions. 18,19 In this study, we investigated whether several viral agents are associated with life-threatening events in hospitalized patients with pneumonia using a case-control study. We found that more than 80% of enrolled hospitalized patients tested positive for respiratory viruses. As expected, RSV was the most frequently observed virus, 3 in children. 23,24 Current clinical guidelines for the management of childhood pneumonia strongly recommend rapid tests for IFV in an inpatient setting. 25,26 Our results further emphasize the critical role of IFV-A infections in severe pneumonia in hospitalized pediatric patients. Early pathogen diagnostics and timely anti-viral therapies would improve mortality. 25,26 We also analyzed the relationships between several host factors and life-threatening events. In our study, almost half of children who had life-threatening pneumonia or who were admitted into the PICU were younger than 6 months of age, and 47% had a pre-existing comorbid condition at the time of hospital admission. Congenital heart disease was the most common