PDF
Citation
Print
INTRODUCTION
Community-acquired lower respiratory tract infection (LRTI), which includes community-acquired pneumonia or bronchiolitis/bronchitis, is a major cause of mortality and morbidity in children.123 The World Health Organization declared that pneumonia was the most common infectious cause of mortality in children worldwide, with 740,180 deaths in children younger than 5 years of age in 2019, accounting for 14% of all deaths in this age group. In addition, bronchiolitis/bronchitis is the most common cause of hospitalisation among infants in developed countries.4
Moderate to severe LRTI in children requires hospitalisation, and antibiotics are required in cases of bacterial or Mycoplasma pneumoniae aetiology.56 Considering that overuse or misuse of antibiotics leads to an increase in antimicrobial resistance, distinguishing whether the LRTI is bacterial, M. pneumoniae, or viral is important.78 However, confirming the pathogen of LRTI is challenging, especially in children owing to difficulty in obtaining adequate specimens, such as a sputum culture.19 Researchers have attempted to differentiate bacterial/M. pneumoniae LRTI from non-bacterial LRTI, using laboratory markers or clinical symptoms, although the application in clinical settings seem to be limited.51011
Considering the large burden of LRTI in children, identifying the epidemiology of LRTIs is important. Therefore, the aim of this study was to describe the demographics, aetiology, clinical manifestation, treatment, and prognosis of children hospitalised with community-acquired LRTI. Moreover, this study aimed to analyse factors that can assist in discriminating the pathogen of LRTI for guidance in proper antibiotic use.
METHODS
Study design, setting, and population
This retrospective cohort study included children younger than 19 years of age who were diagnosed with LRTIs (including pneumonia and bronchiolitis/bronchitis) and required hospitalisation at Seoul National University Bundang Hospital (SNUBH) from January 1, 2005 to July 31, 2019. SNUBH is a 1,335-bed tertiary medical hospital in Gyeonggi-do Province, which is located in the southern region of Seoul (area, 5,919.47 km2), and it is designated as a regional emergency medical institution. Patients with an alternative diagnosis of a respiratory disorder, congenital or neonatal lung disease, aspiration pneumonia, primary malignancy, metabolic disease, or congenital brain malformation; newborns who never left the hospital; and patients who had primary immunodeficiency or had received a solid-organ or hematopoietic stem-cell transplant within the previous 90 days were excluded from this study.
Data collection
Patients were initially identified from the Observational Medical Outcomes Partnership Common Data Model database of SNUBH by chief complaints of fever or acute respiratory symptoms (cough, sputum, dyspnoea, and tachypnoea) and the diagnosis of LRTI (pneumonia, bronchiolitis, and bronchitis). The demographics, underlying disease, vital signs, laboratory findings, microbiology results, treatment, radiologic findings, admission date, duration, and clinical outcome were also collected through the database. To confirm the cases, we retrospectively reviewed the electronic medical records to collect the patients’ symptoms and physical examination findings, and to exclude non-LRTI cases such as upper respiratory tract infections, non-infectious respiratory symptoms (asthma, foreign body aspiration, etc.) and other infections not related to LRTI (sinusitis, tonsillitis, acute otitis media, enteritis, etc.).
Patients were defined as having community-acquired pneumonia if radiologic evidence of pneumonia (consolidation, opacity, infiltration, haziness, or pleural effusion) on chest radiography within 1 week before or after admission was confirmed. Chest radiographs were independently interpreted by board-certified paediatric radiologists.12 Otherwise, the patients were defined as having bronchiolitis/bronchitis. Hospital-acquired pneumonia, which was defined as pneumonia developing 48 hours after hospital admission, was excluded.13
All patients underwent microbiological testing and were also classified based on the microbiological diagnosis. They were classified as having a bacterial infection if Staphylococcus aureus, Streptococcus pneumoniae, Streptococcus pyogenes, viridans group streptococci, Escherichia coli, Haemophilus influenzae, or Klebsiella pneumoniae was detected in the blood, sputum, or pleural fluid. Only sputum cultures obtained in children older than 7 years of age were considered clinically meaningful. Mycoplasma infection was defined in cases in which results of the polymerase chain reaction (PCR) assay for Mycoplasma via a throat swab, nasopharyngeal swab, or lung aspirates were positive; Mycoplasma immunoglobulin (Ig)-M antibodies were detected (> 1.1); at least a 4-fold increase in the mycoplasma IgG antibody titre was detected; or negative to positive mycoplasma IgG antibody conversion was confirmed in paired serologic testing. Patients were classified as having a viral infection if they had positive respiratory virus PCR assay results for respiratory syncytial virus (RSV), influenza, adenovirus, or parainfluenza virus from a throat or nasopharyngeal swab, or if they had a positive antigen assay result for RSV or influenza from a throat or nasopharyngeal swab.
To analyse the clinical manifestations of LRTI, data on the review of systems, laboratory findings, physical examinations, imaging findings, and complications were reviewed. To assess differences, we categorised these manifestations into 2 groups: 1) pneumonia and bronchiolitis/bronchitis and 2) viral pneumonia and bacterial/M. pneumoniae pneumonia. Leucocytosis was defined as a white blood cell count of ≥ 15,000/mm3. Neutrophilia or lymphocytosis was defined when segment neutrophils or lymphocytes constituted > 70% of the total, respectively. Thrombocytosis was defined as a platelet count of ≥ 450,000/mL.
We conducted an analysis comparing bacterial/M. pneumoniae LRTIs to viral LRTIs in relation to season, age, clinical symptoms, laboratory test results, physical examination findings, radiological findings, and adverse effects. Through this analysis, we sought to identify factors that exhibit a stronger correlation with bacterial/M. pneumoniae aetiology.
Statistical analysis
Descriptive statistics were used to compare the demographic characteristics and are presented as counts, percentages, medians, and interquartile ranges (IQRs). Continuous variables were analysed using Student’s t-test, and the χ2 test for independence was used to compare data between each group. Categorical variables were analysed using the χ2 or Fisher’s exact test. To compare features between viral and bacterial/M. pneumoniae LRTI groups, we performed logistic regression analysis. Aetiology-based comparisons were analysed using relative risk (RR) analysis. All data management and statistical analysis were performed using Python (v3.12.1; Python Software Foundation, Wilmington, DE, USA), Pandas (v2.1.4; The Pandas Development Team, https://doi.org/10.5281/zenodo.3509134), and Jupyter (v1.0.0; Jupyter Development Team, https://jupyter.org/about). P values < 0.05 were considered statistically significant in the analyses.
RESULTS
Demographics and patient characteristics
Among 6,367 children, a total of 5,924 cases of community-acquired LRTI were included in this study (Fig. 1). Detailed classifications of the non-LRTI cases are as follows: non-infectious causes of lower respiratory tract symptoms (e.g., asthma, pulmonary artery sling, and foreign body aspiration) (n = 66), croup or laryngitis (n = 42), infectious diseases other than pneumonia or bronchiolitis/bronchitis (e.g., gastroenteritis, febrile convulsion, and septic arthritis) (n = 202), other underlying disease (e.g., primary malignancy, metabolic disease, and congenital malformation) (n = 54), non-infectious disease (n = 35), and unclassified cases (n = 6). Among 5,924 cases, 74.2% (n = 4,393) were cases of pneumonia and 25.8% (n = 1,531) were cases of bronchiolitis/bronchitis. Overall, patients’ median age was 1.8 years (IQR, 3.1 years); median ages of patients with pneumonia and bronchiolitis/bronchitis were 2.5 (IQR, 3.5) and 0.5 (IQR, 0.9) years, respectively (Table 1). Patients with LRTI with underlying disease accounted for 9.5% (n = 565) of the population; chronic lung disease, chronic kidney disease, neuromuscular disease, congenital heart disease, systemic connective tissue disorder, liver disease, obesity, and diabetes mellitus were the most common underlying diseases (Table 1).
Fig. 1
Flow chart of the study population.
OMOP = Observational Medical Outcomes Partnership, LRTI = lower respiratory tract infection.
aAcute respiratory symptoms include cough, sputum, dyspnoea, tachypnoea.
bLRTI includes pneumonia, bronchiolitis, bronchitis.
Table 1
Characteristics of children with community-acquired lower respiratory tract infection
Aetiology
Among the 5,924 LRTI cases, the pathogen was identified in 37.8% (n = 2,242) of cases. Among pathogen identified cases, 69.1% (n = 1,551) were viral, 1.7% (n = 37) were bacterial, and 29.2% (n = 654) were mycoplasma (Table 2). Among 4,393 pneumonia cases, the pathogen was verified in 35.5% (n = 1,558) of cases. Viral pneumonia accounted for 57.2% (n = 891) of pathogen identified cases, bacterial pneumonia for 2.4% (n = 37), and M. pneumoniae for 40.4% (n = 630). Among 1,531 cases of bronchiolitis/bronchitis, 44.7% (n = 684) were pathogen identified. Viral pathogens constituted the majority (96.5%) (n = 660), whereas M. pneumoniae accounted for 3.5% (n = 24). Among viruses, RSV (n = 1,099, 70.9%) was the most common, followed by influenza (n = 210, 13.5%), parainfluenza (n = 187, 12.1%), and adenovirus (n = 114, 7.4%). Among bacteria cases, S. pneumoniae (n = 27, 73.0%) was the most common bacterium, followed by S. aureus (n = 9, 24.3%) and S. pyogenes (n = 1, 2.7%).
Table 2
Aetiology of children with community-acquired lower respiratory tract infection
Clinical manifestations
The clinical manifestations of children with community-acquired LRTI are presented in Table 3. Significant differences between the pneumonia and bronchiolitis/bronchitis groups were observed for several parameters. Fever, sputum, leucocytosis, neutrophilia, C-reactive protein (CRP) elevation, crackles, decreased breath sounds, and positive imaging findings positive imaging findings such as infiltration, consolidation, opacity, haziness and pleural effusion were more indicative of pneumonia. Rhinorrhoea, dyspnoea, lymphocytosis, thrombocytosis, wheezing, chest retraction, and stridor were more indicative of bronchiolitis/bronchitis. Cough was not a significant differentiating factor. Notably, fever was present in 89.8% (n = 3,947) of cases in the pneumonia group compared with 58.3% (n = 893) of cases in the bronchiolitis/bronchitis group (Table 3). Leucocytosis (23.4%, n = 1,026) and CRP level elevation (26.2%, n = 1,152) were higher in the pneumonia group than in the bronchiolitis/bronchitis group (18.2%, n = 279 and 4.3%, n = 66, respectively).
Table 3
Comparison of clinical manifestations between patients with community-acquired pneumonia and bronchiolitis/bronchitis and between bacterial/Mycoplasma pneumoniae pneumonia and viral pneumonia
When comparing bacterial/M. pneumoniae pneumonia to viral pneumonia, fever, neutrophilia, CRP elevation, decreased breath sounds, consolidation, opacity, and pleural effusion were more indicative of bacterial/M. pneumoniae pneumonia. Rhinorrhoea, dyspnoea, lymphocytosis, thrombocytosis, wheezing, chest retraction, and infiltration were more indicative of viral pneumonia. Cough, sputum, leucocytosis, stridor, and haziness were not significant differentiating factors. Although crackles were more indicative of viral pneumonia, the P value was marginal. Patients in the bacterial/M. pneumoniae pneumonia group exhibited fever significantly more often than those in the viral pneumonia group (93.9%, n = 626 vs. 82.6%, n = 736) (Table 3). CRP levels were also notably elevated in the bacterial/M. pneumoniae pneumonia group (30.1%, n = 201) compared with the viral pneumonia group (19.3%, n = 172).
Treatment and outcome
The treatment and outcome of children with community-acquired LRTI are shown in Table 1. The median hospitalisation duration was 4 days (IQR, 2 days). In total, 72.3% (n = 4,283), 82.1% (n = 3,607), and 44.2% (n = 676) of patients with LRTIs, pneumonia, and bronchiolitis/bronchitis, respectively, were treated with antibiotics. In detail, 82.1% (n = 3,607) of patients with pneumonia received antibiotic treatment, while 44.2% (n = 676) of patients with bronchiolitis/bronchitis received antibiotic treatment. Additionally, 97.8% (n = 652) of patients with bacterial/M. pneumoniae pneumonia received antibiotic treatment, and 68.5% (n = 610) of patients with viral pneumonia received antibiotic treatment (Supplementary Table 1). Moreover, 26.5% (n = 1,571), 17.3% (n =7 62), and 52.8% (n = 809) of patients with LRTIs, pneumonia, and bronchiolitis/bronchitis, respectively, received supportive care. Steroids and antiviral agents were used in 4.9% (n = 291) and 0.9% (n = 56) of patients with LRTIs, respectively. Oxygen therapy was administered to 16.1% (n = 952), 12.8% (n = 561), and 25.5% (n = 391) of patients with LRTIs, pneumonia, and bronchiolitis/bronchitis, respectively. Among all patients with LRTIs, 1.3% (n = 75) were admitted to the intensive care unit and 0.9% (n = 56) required mechanical respiratory support. There was no case of mortality.
Analysis of viral LRTI and bacterial/M. pneumoniae LRTI
Results of RR analysis between viral and bacterial/M. pneumoniae LRTIs are displayed in Table 4. Bacterial/M. pneumoniae LRTIs demonstrated a strong association with the summer season (RR, 2.99; 95% confidence interval [CI], 2.42–3.69) and a weak association with the winter season (RR, 0.52; 95% CI, 0.45–0.61). The association with bacterial/M. pneumoniae LRTIs increased with increasing age (RR, 0.10; 95% CI, 0.08–0.13 to RR, 7.06; 95% CI, 4.62–10.78). Fever was associated with bacterial/M. pneumoniae LRTIs (RR, 1.32; 95% CI, 1.27–1.37), whereas cough and sputum had no association. Rhinorrhoea (RR, 0.59; 95% CI, 0.54–0.64) and dyspnoea (RR, 0.44; 95% CI, 0.36–0.55) were associated with viral LRTIs. Bacterial/M. pneumoniae LRTIs showed a strong association with leucocytosis, neutrophilia, and CRP level elevation, whereas viral LRTIs showed a strong association with lymphocytosis and thrombocytosis. Wheezing, stridor, and chest retraction were more commonly observed in viral LRTIs, whereas decreased breath sounds were more common in bacterial/M. pneumoniae LRTIs. Bacterial/M. pneumoniae LRTIs were more commonly associated with imaging findings of consolidation, opacity, haziness, and pleural effusion, whereas viral LRTIs were more commonly associated with infiltration.
Table 4
Results of RR analysis between viral lower respiratory tract infection and antibiotic-requiring lower respiratory tract infection
DISCUSSION
This study was conducted over the longest period and included the largest sample size of single centre study conducted on children in South Korea focusing on community-acquired LRTI requiring hospitalisation. Over a period exceeding 14 years, extensive data collection on paediatric patients was conducted, allowing for a comprehensive exploration of the demographics, aetiology, and clinical manifestations of LRTI cases. Furthermore, these data were used to estimate the aetiology and support decision making for the use or non-use of antibiotics.
According to the overall demographics, the average age was slightly higher in patients with pneumonia than in those with bronchiolitis/bronchitis. Additionally, among children 2 years of age or younger, bronchiolitis/bronchitis had a higher prevalence, as in other studies.1415 Antibiotics were prescribed more than twice as frequently in patients with pneumonia than in those with bronchiolitis/bronchitis, whereas patients with bronchiolitis/bronchitis were treated with supportive care approximately 3 times more often than those with pneumonia. Nevertheless, it was found that antibiotics were still prescribed in approximately half of the cases of viral infections, where antibiotics are not needed (Supplementary Table 1). Oxygen requirements were significantly higher in the bronchiolitis/bronchitis group than in the pneumonia group. RSV is a predominant pathogen in bronchiolitis/bronchitis, and in younger age groups, RSV is reportedly associated with increased oxygen requirements. Our results are consistent with the previous findings.1415
The overall pathogen detection rate was 37.9%, which is slightly lower than that in other studies.16 The majority of cases were attributed to viruses, with M. pneumoniae accounting for 29.2%, whereas the rate of other bacterial detection was notably low. This may be due to the difficulty of obtaining sputum samples in young children. Since diagnostic tests to identify pathogens were performed on all patients, it was found that the diagnostic rate of bacterial pneumonia through blood culture was also very low. Due to the low number of bacterial cases, we were unable to secure sufficient data to observe the effects of changes in the pneumococcal conjugate vaccine (PCV7, 10, 13). Among viruses, RSV was the most common, whereas M. pneumoniae was the most common bacterial pathogen and S. pneumoniae was the second common pathogen. These frequent pathogens were consistent with findings of previous studies.171819 The introduction and expanded indications of RSV antiviral vaccines or monoclonal antibodies will significantly alter the aetiology of viral pneumonia in the future.202122 Although some cases of coinfection were identified, mostly a combination of viruses and M. pneumoniae, the proportion of coinfections was low. Among the 2,242 pathogen identified cases, there were 2 cases (0.1%) of bacterial and viral coinfection, 3 cases (0.1%) of bacterial and M. pneumoniae coinfection, and 42 cases (1.9%) of viral and M. pneumoniae coinfection. Interestingly, evidence of pathogens requiring antibiotics was found in less than half of the patients diagnosed with pneumonia, although most patients in the pneumonia group received antibiotic treatment. This trend has also been observed in other studies.2324
To assist in determining the use of antibiotics, an RR analysis was conducted among cases of identified bacterial/M. pneumoniae LRTI and viral LRTI. Seasonally, LRTIs occurring in the summer tended to be bacterial/M. pneumoniae infections, especially in children aged 2 years or older and particularly in children aged 5 years or older. However, for infections occurring in the winter, the probability of a viral aetiology was significantly higher than in other seasons, as in previous studies (Supplementary Figs. 1 and 2).2526 Especially in Supplementary Fig. 1, viral infections show a significant decrease during July and August, while M. pneumoniae and bacterial infections tend to increase slightly. It is thought that the greater reduction in viral infections during the summer leads to a relative increase in the summer RR ratio for M. pneumoniae and bacteria.
In the review of systems, the presence of fever was associated with a higher likelihood of bacterial/M. pneumoniae infection, whereas symptoms such as cough and sputum production were less decisive in determining antibiotic use. Several studies report using a CRP range of 4 to 6 mg/dL as a cutoff, and higher cutoff values are generally associated with better differentiation between bacterial and non-bacterial aetiologies.2728 Therefore, a CRP level above 5 mg/dL was selected. Laboratory results showed that leucocytosis, neutrophilia, and elevated CRP levels were suggestive of bacterial/M. pneumoniae infection, whereas lymphocytosis and thrombocytosis were more indicative of viral infection. Physical examination findings, such as decreased breath sounds, strongly suggested bacterial/M. pneumoniae infection. Positive imaging findings, especially consolidation, were highly associated with bacterial/M. pneumoniae infection, whereas infiltration on imaging was more suggestive of viral infection. All of the described findings are consistent with those of previous studies.293031 In summary, the factors associated with a higher likelihood of bacterial pneumonia were identified as fever, occurrence during the summer, leucocytosis, neutrophilia, elevated CRP level, decreased breath sounds, and positive imaging findings (consolidation, opacity, haziness, or pleural effusion).
This study had some limitations. First, this was a single-centre study; therefore, it cannot represent the overall incidence of regional or national LRTIs in children. However, the study institution is the largest medical hospital located in the southern region of Gyeonggi-do and serves as a referral centre from primary care centres throughout Gyoenggi-do. Second, we only collected positive PCR results for RSV, influenza, adenovirus, or parainfluenza virus. The wider range of viruses detectable by PCR (human metapneumonovirus, coronavirus, bocavirus, rhinovirus, etc.) could have increased the number of cases classified as pathogen-identified viral pneumonia and viral bronchiolitis/bronchitis. Third, it was difficult to obtain sputum specimens in younger children, and the number of cases classified as bacterial infections was small. This is an important, however common, limitation in diagnosing bacterial pneumonia in children. Fourth, although we attempted to confirm LRTI cases through a detailed chart review, the retrospective nature of medical records inherently limits the clarity of case definitions. For example, children who were hospitalized for the initial manifestation of asthma might have been included in the analysis despite not having LRTI. Finally, the lack of clear pathogen identification is another limitation. Differentiating between colonizers and true pathogens can be challenging, and the sensitivity and specificity of different diagnostic tests (culture, antigen tests, PCR, and serology) can vary, leading to potential false positives and false negatives. Although the prevalence of asymptomatic carriage of M. pneumoniae in Korea is not well known, other reports indicate that it is common in other countries.32 The high outpatient antibiotic prescription rates in Korea may also have influenced the diagnosis of bacterial/M. pneumoniae infections.33 These factors can complicate the accurate determination among recent, past infections or colonization. We acknowledge that these limitations introduce potential biases that could affect our study results.
In children with LRTIs, among cases in which a pathogen was detected, almost 2/3 were due to viral infections, one-third were caused by M. pneumoniae, and few cases were due to other bacterial pathogens. LRTI in children ≥ 2 years of age; summer season; and patients with fever, leucocytosis, neutrophilia, CRP level above 5 mg/dL, decreased breath sounds, and positive image findings (consolidation, opacity, haziness, or pleural effusion) are more likely to have bacterial/M. pneumoniae infections than viral infections. Additionally, a large portion of children may be treated with unnecessary antibiotics. LRTIs in children are a substantial burden with the need for antibiotic stewardship.
XML Download