key: cord-0004960-lnkf9any
authors: Korppi, M.; Heiskanen-Kosma, T.; Jalonen, E.; Saikku, P.; Leinonen, M.; Halonen, P.; Mäkela, P. H.
title: Aetiology of community-acquired pneumonia in children treated in hospital
date: 1993
journal: Eur J Pediatr
DOI: 10.1007/bf02072512
sha: 088e2203dd5a5c1959833865c2c90552355f8b61
doc_id: 4960
cord_uid: lnkf9any

Viral and bacterial antigen and antibody assays were prospectively applied to study the microbial actiology of community-acquired pneumonia in 195 hospitalised children during a surveillance period of 12 months. A viral infection alone was indicated in 37 (19%), a bacterial infection alone in 30 (15%) and a mixed viral-bacterial infection in 32 (16%) patients. Thus, 46% of the 69 patients with viral infection and 52% of the 62 patients with bacterial infection had a mixed viral and bacterial aetiology. Respiratory syncytial virus (RSV) was identified in 52 patients andStreptococcus pneumoniae in 41 patients. The next common agents in order were non-classifiedHaemophilus influenzae (17 cases), adenoviruses (10 cases) andChlamydia species (8 cases). The diagnosis of an RSV infection was based on detecting viral antigen in nasopharyngeal secretions in 79% of the cases. Pneumococcal infections were in most cases identified by antibody assays; in 39% they were indicated by demonstrating pneumococcal antigen in acute phase serum. An alveolar infiltrate was present in 53 (27%) and an interstitial infiltrate in 108 (55%) of the 195 patients. The remaining 34 patients had probable pneumonia. C-reactive protein (CRP), erythrocyte sedimentation rate and total white blood cell count were elevated in 25%, 40% and 36% of the patients, respectively, CRP was more often elevated in patients with bacterial infection alone than in those with viral or mixed viral-bacterial infections. No other correlation was seen between the radiological or laboratory findings and serologically identified viral, bacterial or mixed viralbacterial infections. By using a comprehensive serological panel, the causative agent could be found in over 50% of patients with pneumonia. We conclude that RSV and pneumococcus are the two most common organisms causing pneumonia in children. Our results suggest that mixed viral-bacterial aetiology is common in lower respiratory tract infections affecting children.

Antibody assays are widely used for the aetiological diagnosis of viral or mycoplasmal pneumonia. Moreover, recently developed method for direct detection of viral antigen in samples from the respiratory tract have proven useful in respiratory syncytial virus (RSV), parainfluenza, influenza and adenovirus infections, especially in young children [7, 10, 18] . The assessment of bacterial aetiology, however, is difficult in children with pneumonia or other respiratory syndromes. Less than 5% of bacterial pneumonias in children are bacteraemic with a positive blood culture [4, 33, 39, 48] . Invasive procedures, such as lung puncture, transtracheal aspiration or broncho-alveolar lavation, are justified in severe cases only. Detection of pneumococcal (PNC) or Haemophilus influenzae type b antigens in urine or serum appear to be insensitive indicators of pneumonia when caused by these organisms [12, 39, 40, 50] . Bacterial antibodies to the capsular polysaccharides of PNC have been studied in children with acute otitis media (AOM) [23, 24] and in recipients of pneumococcal vaccine [25, 30] , but only occasionally in children with lower respiratory infections [4, 20, 33] . However, the use of polysaccharide antibody assays may not be suitable, because children under 2 years of age respond poorly to these polySaccharide antigens [25, 30] . Determination of an antibody response to the protein antigen, pneumolysin, has been shown to be useful in adults, but it has been quite insensitive in children with PNC pneumonia [4, 13, 16] . Viral respiratory infection may precede bacterial invasion and infection of the airways [6, 26] . Serological methods have enabled the detection of common mixed viral-bacterial respiratory infections in children [4, 11, 2O-22, 33] .

The aim of the present study was to evaluate the role of viral, bacterial and mixed viral-bacterial aetiology of community-acquired pneumonia in hospitalised children by means of viral and bacterial antibody and antigen assays.

During a surveillance period of 12 months, from 1 September 1981 to 31 August 1982, 195 children (117 less than and 78 over 2 years of age) were treated for pneumonia in the Department of Paediatrics, at Kuopio University Hospital, which provides care for all paediatric patients in a geographically defined area. The average population of children under 2 years of age and under 15 years of age in that area during the study period was 5818 and 45122, respectively. Thus the incidence of community-acquired pneumonia treated in hospital was 20/1000 per year in children under 2 years of age and 4/1000 per year in the whole child population. The diagnosis of pneumonia was based on the presence of pulmonary infiltration as evaluated independently by two radiologists.

The radiographs were later re-evaluated and the findings classified into three groups; definite alveolar or interstitial pneumonia, or probable pneumonia. The diagnosis of AOM was established, if the eardrum was red or cloudy with limited motility.

On admission, acute phase nasopharyngeal (NPS), serum and urine samples were obtained and stored at -20~ until studied (Table 1) . Tests for direct viral antigen detection by radioimmunoassay (RIA) in NPS were performed for RSV, parainfluenza (PV) types 1, 2 and 3, adenoviruses and influenza A and B virus, as previously described [42] [43] [44] . PNC and H. influenzae type b polysaccharide antigens were assayed in acute phase serum samples using a commercially available latex particle agglutination test (Bactigen, Wampole Laboratories, NJ) and PNC antigen also in tenfold concentrated urine by counterimmuno-electrophoresis with PNC Omniserum (Statens Serum Institute, Copenhagen, Denmark). In addition, neutral capsular polysaccharides of PNC types 7 and 14 were detected by a latex particle agglutination test [28] . The concentration of C-reactive protein (CRP; immunonephelometric method), erythrocyte sedimentation rate (ESR; Westergren method) and white blood cell count (WBC) were determined from acute phase blood samples.

Convalescent phase sera were obtained 3-4 weeks later ( Table  2) . Enzyme immunoassay (EIA) was used for measuring pneumococcal C-polysaccharide and type-specific capsular polysaccharide antibodies (the 11 most common PNC types), as well as for measuring antibodies to non-classified H. influenza, MoraxeIla (Branhamella) catarrhalis and Chlamydia species [17, 23, 25, 29, 37] . A threefold or greater rise in IgM, IgG or IgA antibody titres between the paired sera was regarded as significant [17, 20, 33] . The antibody assays were validated by studying paired sera from 200 healthy children and adults; a threefold or greater rise in titre was present in less than 1% between these paired sera [20] . In addition, IgG antibodies to the PNC protein antigen pneumolysin were measured by EIA and a twofold or greater rise was regarded as significant [13, 16] . This assay was validated by studying paired sera from 186 healthy children; a twofold or greater rise in titre was present in 1%-2% between these paired sera. A scaled-down conventional complement fixation (CF) assay was used to measure antibodies to RSV, PV types 1, 2 and 3, adenoviruses, influenza A and B virus, coronaviruses, mumps virus, herpes simplex virus, cytomegalovirus, enteroviruses, M.pneumoniae and the chlamydial group antigen. A fourfold or greater rise in titre was considered significant [9] . The serum pairs showing an antibody rise to Chlamydia species either by EIA or CF were further tested with a micro-immunofluoresence (IF) test specific for C. trachornatis and C. pneumoniae [52] . An infection was considered to have viral aetiology, if any virus was implicated by CF serology, or if any of the seven viral antigens studied was found in NPS by RIA. The patient was considered to have a bacterial infection, if a significant antibody response to any of the five bacteria was demonstrated, or if PNC of H. influenzae type b antigen was found in serum or urine. Infections caused by both viruses and bacteria are called mixed viral-bacterial infections, and those caused by two viruses or two bacteria dual viral or dual bacterial infections.

The chi-square test was used for statistical analysis of the data.

A definite alveolar infiltrate was noted in 27% and a definite interstitial infiltrate in 55% of the 195 patients (Table 3) . Alveolar infiltrates were rare, 20% in infants as opposed to 30%-31% in older children. AOM was seen in 37% of patients younger than 2 years of age and in 24% of the older children (Table 3) . CRP concentrations, ESR levels and WBC counts were elevated in 25%, 40% and 36% of patients, respectively (Table 3) . CRP concentrations and ESR levels were more often elevated in older children than in infants while WBC counts were independent of age. Serological evidence of a viral or bacterial aetiology of an infection was present in 99 (51%) of the 195 children with pneumonia (Table 4) . A viral infection was seen in 37 patients (19%); 35 had a single and 2 a dual viral infection (Table 5) . A mixed viral-bacterial infection was seen in 32 patients; 5 had two bacterial findings. A bacterial infection alone was seen in 30 patients (32%) ( Table 5) . A single bacterial infection was diagnosed in 25 cases and dual bacterial infection in 5 cases (including 1 case with three bacterial findings). Thus, the number of patients with viral infection, either alone or combined with bacterial infection, was 69 (35%). The number of patients with a bacterial infection, whether single, mixed viral-bacterial or dual bacterial, was 62 (32%). RSV was the most common aetiological agent identified in 52 patients (Table 4) . RSV infections were more common in infants than in older children. A single RSV infection was present in 27 cases, a mixed infection with bacteria in 23 cases and a dual infection with adenoviruses in 2 cases (Table 5) . A great majority, 79%, of the 52 RSV findings were obtained by antigen detection tests (Table 6) .

Streptococcus pneumoniae was the second most frequently found aetiological agent identified in 41 patients (Table 4 ). No age dependence was seen. A single PNC infection was present in 19 cases, a mixed infection with viruses in 18 cases and a dual infection with other bacteria in 5 cases (Table 5 ). Over half, 63%, of the 41 PNC infections were found by antibody assays; in 39% the diagnosis was based on the presence of antigen in acute phase serum (Table 6 ). Antibody responses to the protein antigen pneumolysin were found in only 3 cases, whereas responses to type-specific capsular polysaccharide antigens were noted in 15 cases and to C-polysaccharide in 6 cases.

A seroconversion to non-classifiable H. influenzae was demonstrated in 17 patients; all serum and urine a Five cases with one viral and two bacterial findings included b One case with three bacterial findings The frequent occurrence of mixed viral-bacterial infections was an important finding in the present study (Fig. 1) . They were common in all age groups, being observed in 19%-20% of the patients under 2 years, and in 12% of the older children. In fact, such a mixed aetiology was found in 46% of all the viral and in 52% of all the bacterial infections. CRP was more often elevated in patients with bacterial infection alone than in those with viral or mixed viral-bacterial infections. Radiological or other laboratory findings showed no significant correlation with the presence of bacterial, viral or mixed viral-bacterial infections ( Table 7) . This was also seen when the infections caused by specific agents were analysed separately.

The results confirm the importance of viruses, especially RSV, as aetiological agents of pneumonia in children [4, 5, 33, 36] . Serological evidence of a viral infection was seen in 35% of cases; 75% of viral infections were attributable to RSV. In earlier studies, the importance of RSV has been underestimated because antibody assays and viral cultures are insensitive, especially in infants [10, 18, 51] . Accordingly the majority, 78%, of RSV infections in the present study were diagnosed by detecting viral antigen in NPS. To be optimal, the diagnostic panel for respiratory viral infection should include culture, direct antigen detection and IgG antibody assay [32] . Two highly sensitive methods, an immunoperoxidase assay for culture and a time-resolved fluoroimmunoassay for antigen detection have been developed after this study [2, 53] .

Adenoviruses were found in 15% of the viral pneumonias in the present study. These viruses are known to cause severe pneumonias in children with long-lasting airway damage [47] . On the other hand, a cultural finding of adenovirus from a patient with pneumonia may not always be aetiologically associated with the disease, as a latent adenovirus infection may be reactivated by other infections [4] . Antigen detection or seroconversion, which were used in the present study, can be considered as an indication of the aetiology. [19] . All five patients with PV pneumonia in this study had an infection caused by PV types 1 or 3.

The role of bacteria other than M. pneumoniae in the aetiology of pneumonia is difficult to assess. Blood culture gives a definite aetiological diagnosis, but it is positive in less than 5% of children with pneumonia [4, 33, 39, 48] . Bacterial culture or antigen detection in sputum has been useful in adults, but in children these methods indicate the carriage rather than the causative agents of pneumonia [14, 15, 31, 49] . Optimally the determination of bacterial aetiology should be based on the isolation of bacteria from lung tissue. In children with severe pneumonia in developing countries bacteria have been identified from these samples in 79% of cases [46] . In our country, however, lung puncture or other invasive methods are only seldom justified in children with community-acquired pneumonia.

In the present study we measured antibodies in paired sera to three PNC components and to non-classifiable H. influenzae, to M. catarrhalis, to M. pneumoniae and to Chlamydia species. In addition, PNC and H. influenzae type b antigens were assayed in acute blood and urine. By this diagnostic panel, bacterial aetiology was indicated in 32% of children with pneumonia; S. pneumoniae comprised 66% of bacterial infections. The most sensitive methods for the aetiological diagnosis of PNC in the present study were antibody assays to type-specific capsular polysaccharides in paired sera and antigen assays in acute phase serum. Antibody assays are quite insensitive in aetiological diagnosis of bacterial pneumonia, and especially children aged less than 2 years respond poorly to polysaccharide antigens [23, 27, 30] . Antigen detection in urine, as also previously suggested, is an insensitive means to demonstrate PNC aetiology [12, 39, 40, 50] . Demonstration of an antibody response to a PNC protein antigen pneumolysin, although specific and sensitive in adults, was positiv e in only three children [13, 16] . The low sensitivity of pneumolysin antibody assays has been observed in children before and it is possible that antibodies to pneumolysin are bound to circulating immunocomplexes which must be dissociated before testing [4, 45] .

All serum and urine samples were negative for H. influenzae type b antigen. However, antibody responses to non-classified strains were common, 27% of all bacterial findings. Thus, our results support the role of noncapsulated H. influenzae as a real aetiological agent in pneumonia in children [33, 35, 54] . On the contrary, infections caused by M. catarrhalis were rare, 2%.

Antibody responses to Chlamydia species were in eight children, but a micro-IF test verified only one case as C. trachomatis and one as C.pneumoniae infection.

Antibody responses measured by micro-IF are often delayed in primary infection; serum samples should be obtained 6 weeks or more apart [41, 55] . Only one chlamydial infection occurred during infancy, indicating that infantile pneumonia caused by C. trachomatis is uncommon in our country, as in other parts of Europe [4, 37] . Recent serological studies have stressed the importance of C.pneumoniae infections in both developing and industrialised countries [1, 41] . Interestingly, six out of the eight children with chlamydial infection also had AOM.

Both C. trachomatis and C.pneumoniae have been isolated from middle ear fluid samples of patients with AOM [3, 8, 34] . However, culture studies are needed to clarify the ultimate role of C. pneumoniae in pneumonia in children.

A mixed viral-bacterial aetiology was surprisingly common; it was present in 16% of the pneumonic patients of this study. A mixed aetiology constituted 46% of all viral and 52% of all bacterial infections. The frequent occurrence of such mixed infections has been observed in adults with pneumonia [27] and in children with lower respiratory tract infection [11, [20] [21] [22] . As previously suggested, there was no specific viral-bacterial connection, a certain virus preceding a certain bacterial infection; the viral diagnosis does not help clinicians in the selection of antibiotics [20] . Our results showed that neither radiological findings nor laboratory parameters such as CRP, ESR or WBC, although commonly used for this purpose, could discriminate between viral, bacterial or mixed viral-bacterial aetiology of pneumonia [38] .

In conclusion, the study confirms the view that viruses, especially RSV, are common aetiological agents in pneumonia in children. The study also shows that in many patients PNC antigens can be found in acute serum, and that many patients produce seroresponses to these antigens. The sensitivity of all antigen and antibody assays for PNC aetiology, however, appears to be low. Similarly, their correlation with commonly used nonspecific indicators of a bacterial infection as elevated CRP, ESR and WBC is poor. Thus, owing to methodological failures we still underestimate the role of bacteria in pneumonia affecting children. Further studies are needed to improve the non-invasive methods for the diagnosis of bacterial aetiology of pneumonia in children.

Chlamydia pneumoniae, strain TWAR, infection in Seattle children and their families

Laboratory diagnosis of infectious disease; principles and practice

Chlamydia trachomatis in otitis media in children

Etiology community-acquired pneumonia in children based on antibody responses to bacterial and viral antigens

Bronchitis and pneumonia in ambulatory patients

Interaction between viral and bacterial infections in the respiratory tract

Direct detection of viral antigens in respiratory infections by immunoassays: a four year experience and new development

The role of Chlamydia trachomatis in middle ear effusions in children

Diagnostic procedures for viral, rickettsial and chlamydial infections, 5th edn

Antigen detection in the diagnosis of viral infections

Mixed viral and bacterial infections are common in children

Problems in determining the etiology of community-acquired childhood pneumonia

Measurement of antibody responses to pneumolysina promising diagnosis of pneumococcal pneumonia

Diagnosis of pneumococcal pneumonia: a comparison between microscopic examination of expectorates, antigen detection and cultural procedures

Diagnosis of pneumococcal pneumonia by coagglutination and counterimmunoelectrophoresis of sputum samples

Granstr0m M611by R (1988) Serum antibodies to pneumolysin in patients with pneumonia

The aetiology of pneumonia, Application of bacterial serology and basic laboratory methods

Viral findings in children under the age of two years with expiratory difficulties

The role of parainfluenza viruses in inspiratory difficulties in children

Bacterial coinfection in children hospitalized with respiratory syncytial virus infections

Bacterial involvement in parainfluenza virus infection in children

Mixed infection is common in children with respiratory adenovirus infection

Serum antibodies to pneumococcal C-polysaccharide in children: a response to acute pneumococcal otitis media or to vaccination

Serum antibody response to pneumococcal otitis media

First and second dose antibody responses to pneumococcal polysaccaride vaccine in infants

Editorial (1982) How does influenza pave the way for bacteria? Lancet I

Etiological diagnosis of pneumonia in military conscripts by combined use of bacterial culture and serological methods

Detection of pneumococcal capsular polysaccharide antigens by latex agglutination, counterimmunoelectrophoresis, and radioimmunoassay in middle ear exudates in acute otitis media

Preliminary serologic evidence for a pathogenic role of Branhamella catarrhalis

Antibody response to 14-valent pneumococcal capsular polysaccharide vaccine in preschool age children

Occurrence of DipIococcus pneumoniae in the upper respiratory tract of children

Immunoglobulin class-specific antibody response in respiratory syncytial virus infection measured by enzyme immunoassay

The etiology of hospitaltreated acute lower respiratory tract infection in children

Isolation of chlamydia trachomatis from the middle ear aspirates of otitis media

Aetiology of pneumonia in children in Goroka Hospital

The child with pneumonia: diagnostic and therapeutic considerations

Chlamydial pneumonitis and its serodiagnosis in infants

C-reactive protein in the evaluation of febrile illness

Use of bacterial antigen detection in the diagnosis of pediatric lower respiratory tract infections

Counterimmunoelectrophoresis and latex particle agglutination in the etiologic diagnosis of presumed bacterial pneumonia in pediatric patients

Acute lower-respiratory-tract infection associated with chlamydial TWAR antibody in Filipino children

Detection of respiratory syncytial, parainfluenza type 2, and adenovirus antigens by radioimmunoassay and enzyme immunoassay in nasopharyngeal specimens from children with acute respiratory disease

Detection of influenza A virus by radioimmunoassay and enzyme immunoassay from nasopharyngeal specimens

Type specific detection of parainfluenza viruses by enzyme immunoassay and radioimmunoassay in nasopharyngeal specimens of patients with acute respiratory disease

Sequestration of antibodies to BorreIia burgdorferi in immune complexes in seronegative Lyme disease

Diagnosis of acute bacterial pneumonia in Nigerian Children

Chronic lung damage caused by adenovirus type 7: a ten-year follow-up study

Bacteremia in febrile children under 2 years of age: Results of cultures of blood in 600 consecutive children seen in a "walk-in" clinic

Turnover of nontypable Haemophilus iufluenzae in the nasopharynges of healthy children

Counterimmunoelectrophoresis of urine for diagnosis of bacterial pneumonia in pediatric patients

Age-specific prevalence of complement-fixing antibodies to sixteen viral antigens: a computer analysis of 58.500 patients covering a period of eight years

Immunologic relationship between genital TRIC, lymphogranuloma venereum, and related organisms in a new microtiter immuno-fluorence test

Rapid detection of respiratory syncytial virus and influenza A virus in cell cultures by immunoperoxidase staining with monoclonal antibodies

Clonal analy-Sis of Haemophilus influenzae isolated from children from Pakistan with lower respiratory tract infections

Transmission of Chlamydia pneumoniae in young children in Japanese family

Acknowledgements. The authors wish to thank Seppo Soimakallio, MD, and Olavi Kiekara, MD, for the interpretation of the radiological findings, Pirjo Halonen, MSc, for statistical advice and Manu Munter, MSc, for the CRP determinations. The study was financially supported by grants from the Academy of Finland (MK, ML) and from the Foundation for Paediatrics Research, Finland.