Helsinki University Biomedical Dissertations No.7

Hospital for Children and Adolescents, Helsinki University Central Hospital,

Helsinki, Finland

Etiology and Antimicrobial Treatment ofPneumonia and Other Common Childhood

Infections Warranting Hospitalization

Elina Vuori-Holopainen

Academic Dissertation

To be publicly discussed with the permission of the Medical Faculty of theUniversity of Helsinki, in the Niilo Hallman Auditorium of the Hospital for

Children and Adolescents,on September 28th 2001, at 12 noon.

Helsinki 2001

2

Supervised byProfessor Heikki Peltola, MDHospital for Children and AdolescentsHelsinki University Central HospitalHelsinki, Finland

Reviewed byProfessor Jussi Mertsola, MDDepartment of PediatricsTurku University HospitalTurku, FinlandandProfessor Matti Korppi, MDDepartment of PediatricsUniversity of KuopioKuopio, Finland

Dissertation opponentDocent Per Ashorn, MDMedical SchoolUniversity of TampereTampere, Finland

ISSN 1457-8433ISBN (nid.) 952-10-0117-8ISBN (pdf) 952-10-0118-6http://ethesis.helsinki.fi

Layout: Callide/Terttu HalmeYliopistopaino, Helsinki 2001

3

Kuka tietävi mistä me tulemmeJa missä on matkamme määrä?Hyvä, että me sitäkin tutkimme.Ei tutkimus ole väärä.Mut yhden me tiedämme varmaan vaan:Me olemme kerran nyt päällä maanja täällä meidän on eläminenmiten taidamme parhaiten.- Eino Leino -

To Juha and Ronja

4

5

CONTENTS

1 ABSTRACT 9

2 ABBREVIATIONS 11

3 LIST OF ORIGINAL PUBLICATIONS 12

4 INTRODUCTION 13

5 REVIEW OF THE LITERATURE 145.1. Definition of lower respiratory infections 145.2. Epidemiology and pathogenesis of lower respiratory infections 14

5.2.1. Epidemiology 145.2.2. Pathogenesis 15

5.3. Pathogens associated with lower respiratory infections 165.3.1. Streptococcus pneumoniae 20

5.3.1.1. Serotypes and virulence factors 205.3.1.2. Carriage 215.3.1.3. Pneumococcal infections 215.3.1.4. Prevention of pneumococcal infection 22

5.3.2. Haemophilus influenzae 225.3.3. Moraxella catarrhalis 235.3.4. Mycoplasma pneumoniae 235.3.5. Chlamydia pneumoniae 245.3.6. Respiratory viruses 245.3.7. Mixed infections 25

5.4. Methods for identification of etiology 265.4.1. Bacterial agents 27

5.4.1.1. Culture 275.4.1.2. Antigen detection 275.4.1.3. Antibody and immune complex assays 28

5.4.1.3.1. Pneumococcal serology 285.4.1.3.2. Other bacterial serology 29

5.4.1.4. Nucleic acid detection 305.4.2. Viral agents 315.4.3. Lung tap in childhood pneumonia 32

5.4.3.1. Earlier clinical experience 325.4.3.2. Risks of the procedure 32

6

5.5. Acute host response parameters distinguishing viral from 33bacterial etiology

5.6. Antimicrobial therapy 345.6.1. Treatment in relation to antimicrobial resistance 35

5.6.1.1. Streptococcus pneumoniae 355.6.1.2. Influence of β-lactamase production 36

5.6.2. Antimicrobial treatment studies 365.6.3. Length of antimicrobial treatment 375.6.4. Current guidelines for treatment of pneumonia 38

6 AIMS OF THE STUDY 386.1. General aim 386.2. Specific aims 38

7 PATIENTS AND METHODS 397.1. SE-TU study (I-III) 39

7.1.1. Study design 397.2. KE-TU study (IV, V) 41

7.2.1. Study design 417.2.2. Technique of lung tap 42

7.3. Specimens and microbiologic methods 427.3.1. Conventional microbiology (I-V) 427.3.2. Serology (I-III) 427.3.3. Lung aspirates (IV, V) 447.3.4. Serotyping and susceptibility testing of Streptococcus 44

pneumoniae (V)7.4. Criteria for etiology (I-V) 457.5. Radiologic assessment (I-V) 467.6. Statistical methods (II, III, V) 46

8 RESULTS 478.1. Etiology in the four clinical manifestations (I) 478.2. Patients with randomized treatment (II, III) 49

8.2.1. Characteristics and bacterial etiology (II, III) 498.2.2. Antimicrobial therapy, length of treatment, 52

and recovery (II, III)8.2.3. Clinical outcome (II, III) 54

7

8.3. Lung tap (IV, V) 558.3.1. Sample validity and overall yield 558.3.2. Etiology of consolidated pneumonia 55

8.3.2.1. Moraxella osloensis (IV) 568.3.3. Adverse events 57

8.4. Diagnostics of pneumococcal pneumonia with 61different methods (I, V)

8.5. Pneumococcal serotypes and antimicrobial susceptibility (V) 62

9 DISCUSSION 639.1. Etiology of pneumonia and other suspected invasive 63

infections warranting hospitalization (I)9.1.1. Study design 639.1.2. Microbiologic methods 649.1.3. Etiologic findings in relation to previous studies 65

9.2. Effectiveness of antimicrobial treatment (II) 669.3. Shortening of therapy (III) 689.4. Lung tap in the diagnosis of pediatric pneumonia (IV, V) 69

9.4.1. Representativeness of the sample 699.4.2. Positivity rate of lung tap 699.4.3. Ethical considerations and applicability of lung tap 71

9.5. Invasive isolates of Streptococcus pneumoniae (V) 729.6. General discussion and future considerations 72

10 SUMMARY AND CONCLUSIONS 74

11 ACKNOWLEDGMENTS 75

12 REFERENCES 77

ORIGINAL PAPERS 89

8

9

1 ABSTRACT

Background: An antimicrobial for suspected bacterial pneumonia and other commonchildhood infections is generally chosen without knowledge of the causative pathogen.The etiologic diagnosis of pneumonia is difficult, because uncontaminated samplesare not easily obtained and indirect, rapid, and accurate laboratory tests for mostpathogens are not available. More sensitive and specific diagnostic methods areneeded. The aim of the present study was to investigate the etiologic diagnostics andtreatment of common invasive bacterial infections of children, with special emphasison alveolar pneumonia.Patients and methods: The effectiveness of two parenteral antimicrobialregimens (procaine penicillin i.m. vs. cefuroxime i.v.) and two different lengths(4 days vs. 7 days) of treatment were compared in a prospective, randomizedstudy comprising 154 3-month-old children with symptoms and signs suggestive ofacute invasive bacterial infection. The patients were enrolled during two time periodsfrom February 1988 to October 1989, and from March 1991 to October 1993.The etiology was studied with oropharyngeal and blood cultures, viral antigen detectiontests, and serology including antibody assays of 15 viruses, seven bacteria, and oneprotozoan.

Microbiologic diagnosis of childhood pneumonia was further investigatedby performing a lung tap on 34 children with consolidated community-acquiredpneumonia during a three-year study period from December 1997 to December2000. In addition to conventional cultures, nucleic acid detection was performedon the lung aspirates for the common respiratory pathogens. The yield obtainedby lung tap was compared with that obtained by routine noninvasive microbiologicmethods. Antimicrobial susceptibility and serotypes of Streptococcus pneumoniaeisolates were studied.Results: The results of this study confirm the importance of S. pneumoniae asa causative bacterial agent in common childhood infections. Serologic evidencefor a pneumococcal etiology was found in 46 % of patients with pneumonia, in30 % of those with other respiratory infection, in 24 % of patients with sepsis-like infection, and in 33% of those with other acute infection. In the treatmentof these infections, no significant difference was observed in the clinical efficacyof parenteral penicillin as compared with cefuroxime. Recovery was similarregardless of the regimen used, there being one potential failure in each treatmentgroup, and no difference in the frequency of needing a physician within 1 monthafter discharge. The outcome of patients treated with 7-day course of β-lactamantimicrobials was not better than that of patients treated for 4 days. There wasone clinical failure potentially related to the shorter course of treatment.

The important role of S. pneumoniae as a cause of pneumonia was furtherconfirmed by studies of lung aspirates, in which it was found in 53 % (18/34) of

10

patients. In all, lung tap provided the etiologic diagnosis in 59% (20/34) ofcases. Two patients had dual bacterial infection, one had mixed bacterial-viralinfection and one had single viral infection. The procedure did not cause anyadverse events requiring treatment.Conclusions: The results of this study suggest that in lower respiratory infectionsand in many other bacterial infections in children, short courses of narrow-spectrum parenteral antimicrobials are justified. In studies of the etiology ofchildhood consolidated pneumonia, lung taps provide specific data with arelatively low rate of adverse events.

11

2 ABBREVIATIONS

CF complement fixation

Cfm cefuroxime

C-PS C-polysaccharide

CRP C-reactive protein

EIA enzyme immunoassay

ESR erythrocyte sedimentation rate

Hib Haemophilus influenzae type b

Ig immunoglobulin

IC immune complex

i.m. intramuscularly

i.v. intravenously

MIC minimal inhibitory concentration

MIF micro-IF, microimmunofluorescence

NPA nasopharyngeal aspirate

PCR polymerase chain reaction

Pen procaine penicillin

Ply pneumolysin

RSV respiratory syncytial virus

WBC white blood cell count

12

3 LIST OF ORIGINAL PUBLICATIONS

This thesis is based on the following articles, which are referred to in the textwith roman numerals (I-V).

I Vuori E, Peltola H, Kallio MJT, Leinonen M, Hedman K, SE-TU StudyGroup. Etiology of pneumonia and other common childhood infectionsrequiring hospitalization and parenteral antimicrobial therapy. Clin InfectDis 1998;27:566-572.

II Vuori-Holopainen E, Peltola H, Kallio MJT, SE-TU Study Group. Narrow-versus broad-spectrum parenteral antimicrobials against common infectionsof childhood: a prospective and randomised comparison between penicillinand cefuroxime. Eur J Pediatr 2000;159:878-884.

III Peltola H, Vuori-Holopainen E, Kallio MJT, SE-TU Study Group. Successfulshortening from seven to four days of parenteral beta-lactam treatment forcommon childhood infections: A prospective and randomized study. Int JInfect Dis 2001;5:3-8.

IV Vuori-Holopainen E, Salo E, Saxén H, Vaara M, Tarkka E, Peltola H. Clinical“pneumococcal pneumonia” due to Moraxella osloensis: Case report and areview. Scand J Infect Dis 2001;33:625-627.

V Vuori-Holopainen E, Salo E, Saxén H, Hedman K, Hyypiä T, Lahdenperä R,Leinonen M, Tarkka E, Vaara M, Peltola H. Etiological diagnosis ofchildhood pneumonia by lung tap, applying modern microbiology. Submitted.

The thesis also includes some previously unpublished data.

13

4 INTRODUCTION

In nonindustrialized countries, the most frequent causes of morbidity andmortality among children are infectious diseases, especially lower respiratoryinfections (Berman 1991, Pechére 1995). Annually, about three million childrendie from pneumonia (Berman 1991). In industrialized countries, lower respiratoryinfections are also a significant cause of morbidity and hospitalization in children< 5 years of age (Murphy et al. 1981, Jokinen et al. 1993, Ruuskanen andMertsola 1999).

Despite the frequency of these diseases, the causes of pneumonia and othercommon invasive bacterial infections are seldom identified in clinical practice.In Europe, knowledge of the etiology of pneumonia has mostly been based onserology and antigen detection methods (Ruuskanen and Mertsola 1999). Withfew exceptions, serology requires paired serum samples, and the test results areonly available after several weeks. Culture-based methods are more rapid, butwith the exception of blood cultures, which are positive in only 0 to 5 % ofchildren requiring hospitalization (Nohynek et al. 1991, Clements and Stephenson1996, Genrel et al. 1997, Juvén et al. 2000), they are frequently complicated bythe difficulty of obtaining accurate specimens. Bacteria detected in upper airwaysamples only demonstrate carriage, and often have little or no association withthe infection in the lower airways (Cunanan et al. 1979, Korppi et al. 1992b). Innonbacteremic cases, convincing evidence of etiology is obtained only byidentification of an agent from the pleural fluid or lung parenchyma (Finland1969, Ruuskanen and Mertsola 1999).

The low diagnostic yield has contributed to the frequent and inappropriateuse of broad-spectrum antimicrobials in childhood infections (Clements et al.2000). However, the use of antimicrobials should be restricted, since the amountsprescribed, their antimicrobial spectra, and the lengths of treatment are strongpredictors of the development and spread of resistant micro-organisms (Arasonet al. 1996, Goldmann et al. 1996, Seppälä et al. 1997, Swartz 1997, Guillemotet al. 1998). Consequently, in addition to etiologic data, clinical trials of theefficacy of an antimicrobial regimen are needed to establish recommendationsregarding appropriate treatment in different clinical situations (Berner 2000).

In the present study, the effectiveness of two antimicrobial regimens and twodifferent lengths of treatment were compared in a prospective, randomized seriesof children with signs and symptoms suggestive of an acute invasive bacterialinfection. The etiology was investigated both by conventional methods and byserology. In a further study of childhood community-acquired pneumonia, anotherapproach was taken. Lung tap (transthoracic needle aspiration) was performedon hospitalized patients with consolidated pneumonia. The use and limitationsof the lung tap were elucidated by comparing the tap yields with those obtainedby other routine methods.

14

5 REVIEW OF THE LITERATURE

5.1. Definition of lower respiratory infections

Pneumonia is defined as inflammation and consolidation of lung tissue due to aninfectious agent (Marrie 1994, Jadavji et al. 1997). The clinical symptoms andsigns are nonspecific and variable (Jadavji et al. 1997, Tan et al. 1998), andtherefore, in industrialized countries, radiologic confirmation of pneumonia isconsidered the gold standard (Marrie 1994, Bartlett and Mundy 1995, Jadavjiet al. 1997). In nonindustrialized countries, radiographic facilities are so seldomavailable that the World Health Organization defines moderate to severe pneumoniain those circumstances as an illness with cough or difficulty in breathing, increasedrespiratory rate and chest indrawing (WHO 1994, Shann 1995).

In community-acquired pneumonia, the infection has been acquired outsidethe hospital, and there are two main classic presentations (Jadavji et al. 1997):1) typical pneumonia with acute onset, dominated by symptoms of fever, chills,pleuritic chest pain, and a productive cough, and 2) atypical pneumonia withgradual onset over several days to weeks, dominated by symptoms of headacheand malaise, nonproductive cough, and low-grade fever. However, in manycases the clinical manifestations differ from these classic presentations, andthe etiologic agent cannot be identified on the basis of the clinical symptomsand signs.

Radiographically, pneumonia has been divided traditionally into twomanifestations: interstitial and alveolar. Diffuse interstitial infiltrates tend to beseen in viral infections (Jadavji et al. 1997), while lobar consolidation suggestsa bacterial etiology. Alveolar infiltrates are seen in both bacterial and viralpneumonia (Jadavji et al. 1997). On chest radiograph, the size and shape of theinfiltration change during the course of infection and influence the radiographicsigns. The value of chest radiographs in differentiating bacterial from viral lowerrespiratory infections in children was recently reviewed (Swingler 2000), andno clinically relevant accuracy was demonstrated.

5.2. Epidemiology and pathogenesis of lowerrespiratory infections

5.2.1. Epidemiology

The overall incidence of acute respiratory infections is very similar in industrializedand in nonindustrialized countries (Berman and McIntosh 1985), but there aremarked differences in their severity. The incidence of lower respiratory infections in

15

the nonindustrialized world is as much as 10 times higher than in industrializedcountries, being in some areas 300/1 000/year among children < 5 years ofage (McCracken 2000). In the Finnish population, in contrast, the overallincidence of community-acquired pneumonia is 37/1 000/year among children< 5 years of age, and 16/1 000/year among those between 5 and 14 years(Jokinen et al. 1993). In the USA, the incidence of pneumonia in children < 6years of age was 30/1 000/year in 1966-1971 (Foy et al. 1973). In general,the incidence varies with the social status, nutritional state, passive exposureto smoke, and attendance at a day-care center (Jadavji et al. 1997). However,in a recent Finnish population based analysis the most important risk factorfor community-acquired pneumonia in children was susceptibility to respiratoryinfections (Heiskanen-Kosma et al. 1997).

Not only the incidence, but also the mortality from pneumonia is higher inthe developing countries, mostly on account of the greater severity of the disease.Worldwide, pneumonia is responsible for a quarter of all deaths in children < 5years of age (Berman 1991). In industrialized countries, the mortality rate hasdeclined markedly in recent decades (Dowell et al. 2000). In the USA, the casefatality of pneumonia in children has declined by 97 %, from 24,637 cases in1939 to 800 in 1996 (Dowell et al. 2000). In Finland, three children died fromcommunity-acquired pneumonia in 1994-1995 (Ruuskanen and Mertsola 1999).In our circumstances, approximately half of the < 5-year-old children withpneumonia are admitted to hospital, whereas most of the older children aretreated at home (Jokinen et al. 1993). In future, vaccines against S. pneumoniaeand influenza virus may offer a chance to reduce the incidence of pneumonia inboth the industrialized and the nonindustrialized world (King et al. 1998, Rubin2000), as has already occurred for Haemophilus influenzae type b (Hib)pneumonia in the industrialized countries as a result of the introduction of generalvaccinations (Peltola 2000).

5.2.2. Pathogenesis

Bacteria causing pneumonia may gain access to the lung by aspiration of mucuscontaminated by the normal oropharyngeal flora. The pathogen’s ability to adhereto the respiratory tract epithelium and to overcome the host’s defense mechanismsis crucial for infectivity (Bakaletz 1995). In most cases, adhesion depends onepithelial damage, which may be caused by a concomitant or previous upperrespiratory viral infection (Fainstein et al. 1980). In addition to damaging themucosal cells, viruses lead to pulmonary infection by causing increased mucusproduction, decreased ciliary action, epithelial swelling and dysfunction ofleukocytes (Berendt et al. 1975, Carson et al. 1985).

16

In acute otitis media, evidence for interactions between bacteria and viruseshas been obtained from both epidemiologic and microbiologic studies(Hendersson et al. 1982, Ruuskanen et al. 1989). The peak incidence of theviral illness occurs either concurrently with or immediately preceding that ofthe bacterial disease (Ruuskanen et al. 1989). Viral infections in the nasopharynxseem to relate more directly to the development of otitis media thannasopharyngeal colonization with S. pneumoniae or H. influenzae alone(Hendersson et al. 1982). In children with invasive pneumococcal disease, theassociation with viral respiratory infection is deemed less apparent and lessimmediate, the lag period being approximately four weeks (Kim et al. 1996).However, previous influenza infection has been demonstrated to predispose tosevere pneumococcal pneumonia (O’Brien et al. 2000).

The outcome of infection depends on the balance between the host defensemechanisms and the growth rate of the invading pathogen (Johnston 1991). In thelungs, alveolar macrophages and polymorphonuclear leukocytes clear the bacteria,provided that specific antibodies and complement are present. In the absence ofspecific antibodies, clearance of bacteria may be enhanced by C-reactive protein(CRP) -mediated complement activation (Johnston 1991, Jarva et al. 1999).

5.3. Pathogens associated with lowerrespiratory infections

A wide range of pathogens has been associated with lower respiratory infections.The spectrum of etiologic agents in industrialized countries resembles that innonindustrialized countries (Table 1), albeit pathogens such as Staphylococcusaureus account for a higher proportion of cases in developing countries. In theindustrialized world, viruses cause a high proportion of cases. According to theculture-based studies, the predominant bacteria causing childhood lowerrespiratory infections worldwide are pneumococci along with H. influenzae(Klein 1981, Shann 1986, Vuori-Holopainen and Peltola 2001). However, theetiology of pneumonia is a complex issue, and many cases remain without anidentified etiology. Some series from nonindustrialized countries have reportedup to 15-45 % of blood cultures as positive (Fagbule 1993, Adegbola et al.1994), whereas, in industrialized countries, positive results are obtained in onlyapproximately 0-5 % of cases (Paisley et al. 1984, Nohynek et al. 1991, Genrelet al. 1997, Juvén et al. 2000). Sometimes, different organisms have been detectedsimultaneously from blood and lung (Shann et al. 1984, Falade et al. 1997); infact, bacteremia has been suggested to be secondary to initially non-bacteremicpneumonia (Scott and Hall 1999). This may explain the higher incidence ofpositive blood cultures in nonindustrialized countries.

17

* Only some of the representative studies included. If several methods were used, only serologic results were accepted.

** Percentage counted from patients in whom the specific agent was searched for by serology.

***No data

According to a recent review of lung tap in childhood pneumonia, a pathogen wasidentified in 32 % to 66 % of patients from different continents, even though inmost cases standard culture techniques only were used (Vuori-Holopainen andPeltola 2001). However, very few data were available from the western world.Most etiologic information from developed countries has been derived either fromretrospective analyses or from studies of hospitalized patients. The results of suchstudies are difficult to compare because of different inclusion criteria and diagnosticmethods. Table 2 presents a summary of etiologic studies from developed countries.

Table 1. Overall frequency of the bacterial agents in studies of pneumonia in childrenconducted since 1980.

Lung tap studies

Continent Lung tap S. pneumoniae H. influenzae M. catarrhalis S. aureus E. coli Klebsiella spp Others ReferenceN N % N % N % N % N % N % N %

Africa 462 61 13 13 3 1 0.2 83 18 17 4 55 12 44 10 Vuori-Holopainenet al. 2001

Asia 292 29 10 4 1.4 1 0.3 40 14 2 1 12 4 36 12 Vuori-Holopainenet al. 2001

Oceania 101 34 34 42 42 9 9 1 1 0 0 0 0 23 23 Vuori-Holopainenet al. 2001

Serology-based studies*

Continent Patients S. pneumoniae H. influenzae M. catarrhalis S. aureus E. coli Klebsiella spp Others

Africa 164 14 9** 13 8 9 5 ND*** ND ND 15 9 Forgie et al.1991a, 1991b

Asia 100 0 0 9 9 1 1 ND ND ND 30 30 Yang et al. 2000

North 168 35 27 ND ND ND ND ND 22 13 Wubbel et al.1999America

Europe 1157 244 21 76 8 31 3 ND ND ND 118 10 Claesson et al.1989,Ruuskanen et. al.1992, Korppi et al.1993a, Nohyneket al. 1995,Heiskanen-Kosmaet al.1998,Juven et al. 2000

18

Table 2. Studies from developed countries in which both viral and bacterialetiology (%) in childhood pneumonia were searched for.

Country N % of patients No of microbes Etiology Bacterial Mixed viral- Viralhospitalized studied detected bacterial

USA 102 100 16 85 11 28 47

USA 162 20 17 48 22 8 19

USA 98 21 10 48 9 10 39

USA 168 0 13 43 25 3 15

France 104 50 12 84 47 8 29

Spain 198 19 11 79 58** ND 27**

Sweden 336 50 13 48 19 4 25

Finland 121 100 15 70 25 20 25

Finland 50 100 16 88 28 34 26

Finland 195 100 14 50 15 16 19

Finland 201 32 13 66 41 10 15

Finland 254 100 15 85 22 30 32

mean % 66 22 16 27

* Pnc, S. pneumoniae; Hi, nontypable H. influenzae; Hib, H. influenzae type b; M. cat, M. catarrhalis; M. pn, M. pneumoniae.

** Two pathogens from 11 children, not specified whether two bacteria, two viruses, or mixed viral-bacterial infections.

19

Table 2. Continued.

Country N Pnc* Hi/Hib* M.cat* M. pn* RSV Adeno Influenza Parainfluenza Rhino Referencevirus A or B 1, 2, or 3 virus

USA 102 23 2/3 ND ND 60 2 1 5 5 Paisley et al. 1984

USA 162 18 ND/7 ND 8 9 4 6 6 4 Ramsey et al. 1986

USA 98 17 ND/2 ND 0 28 1 2 5 ND Turner et al. 1987

USA 168 27 ND/ND ND 7 8 2 3 4 ND Wubbel et al. 1999

France 104 13 ND/ND ND 41 10 4 4 6 ND Genrlel et al. 1997

Spain 198 15 ND/2 ND 40 15 3 4 5 ND Ausina et al. 1988

Sweden 336 13 ND/1 ND 10 20 5 3 3 ND Claesson et al.1989

Finland 121 16 16/2 7 9 28 13 3 7 ND Nohynek et al.1991

Finland 50 38 12/ND 10 20 30 10 2 8 10 Ruuskanenet al.1992

Finland 195 21 21/ND 2 2 27 5 ND 3 ND Korppi et al. 1993a

Finland 201 28 6/ND 3 22 21 2 0 3 ND Heiskanen-Kosmaet al.1998

Finland 254 37 9/ND 4 7 29 7 4 10 24 Juvén et al. 2000

mean % 22 11/3 6 16 23 5 3 6 11

* Pnc, S. pneumoniae; Hi, nontypable H. influenzae; Hib, H. influenzae type b; M. cat, M. catarrhalis; M. pn, M. pneumoniae.

** Two pathogens from 11 children, not specified whether two bacteria, two viruses, or mixed viral-bacterial infections.

20

5.3.1. Streptococcus pneumoniae

5.3.1.1. Serotypes and virulence factors

S. pneumoniae is enveloped by a polysaccharide capsule determining the serotypeof each pneumococcus. According to the Danish nomenclature (Forward 1999),90 serotypes have been recognized to date. Based on information obtained mostlyfrom otitis media, meningitis, and blood cultures, the serogroups currentlyconsidered to be the most important among pediatric patients in industrializedcountries are 1, 4, 6, 7, 9, 14, 15, 18, 19, and 23 (Eskola et al. 1992, Butler etal. 1995, Sniadack et al. 1995, Toikka et al. 1999b). Recently, in a Finnishanalysis of otitis media most common serogroups were found to be 6, 11, 15,18, 19, and 23 (Eskola et al. 2001). Since childhood pneumonia is rarelybacteremic and lung aspirates are seldom obtained, only limited data onpneumococcal serogroups in pneumonia are available (Table 3).

Besides capsular polysaccharides, pneumococci have several other virulencefactors, such as C-polysaccharides (C-PS) and pneumolysin (Ply), a toxicintracellular protein, both of which are common to all strains (Tuomanen et al.1987, Paton et al. 1993). Pneumococcal surface protein A and surface adhesinA are also found in most clinical isolates, and these components, too, maycontribute to the pathogenicity of S. pneumoniae (Briles et al. 1988, Rapola etal. 2000). In pneumococcal infections, antibody responses may be detected toall these virulence factors.

Source of Number of Country Age Serogroups** Referenceisolates* isolates

LA 7 Papua New Guinea ND*** 5, 6, 7, 14, 16, 46 Riley et al. 1983

LA, B 25 Papua New Guinea 1m-4y 6, 14, 19, 5, 16, 23, 4, 12, 15, 17, 22, 25, 45 Shann et al.1984

LA, B, PF 46 Gambia 3m-5y 6, 14, 5, 1, 9, 12, 13, 19 Adegbola et al.1994

LA, B 14 Gambia 2m-10y 1, 5, 6, 14, 19 O’Neill et al.1989

B 61 Finland ≤15y 7, 19, 14, 6, 23, 9, 4, 3, 10, 38 Toikka et al.1999b

* LA, lung aspirate; B, blood; PF, pleural fluid.

** Presented in descending order of prevalence; numbers of serotypes found in >10 % of isolates bolded.

*** ND, no data.

Table 3. Pneumococcal serogroups recovered from childhood pneumonia.

21

5.3.1.2. Carriage

Pneumococci colonize the human pharynx commonly and asymptomatically;> 95 % of children carry pneumococcus at some time during the first 2 years oflife (Klugman 1990). The way of potential translocation of pneumococci fromthe nasopharynx to the middle ear, lungs, blood, or meninges is not wellunderstood. In most cases, symptomatic infection arises from a recently acquiredserotype (Gray et al. 1980). Acquisition of a new serotype leads to infection in15 % of cases, and approximately 75 % of pneumococcal infections occur withinone month after acquisition (Gray et al. 1980). In 90 % of children with an invasivedisease, pneumococcus has been isolated from the nasopharynx, the strainsbelonging to the same serotype in 77 % of cases (Lloyd-Evans et al. 1996).

5.3.1.3. Pneumococcal infections

S. pneumoniae is considered to be the most frequent bacterial cause ofpneumonia, otitis media, sinusitis, and bacteremia. In areas where Hib has beenvirtually eliminated by vaccine, it is also often the main pathogen in childhoodmeningitis (Klein 1981, Musher 1992). According to serology-based studies,pneumococcus is, with few exceptions (Yang et al. 2000), the most commoncausative agent of childhood pneumonia in both industrialized and non-industrialized countries (Tables 1 and 2). In North America and in Europe, it isfound in 18-27 % and 13-38 % of children with mainly hospitalized pneumonia,respectively. The full role of pneumococcus is difficult to assess, since all availableassays for optimal detection of pneumococci have seldom been used (Heiska-nen-Kosma 2000). In lung tap studies on 855 pediatric pneumonia patientsconducted since 1980, a pneumococcal etiology has been confirmed in only 10 %of Asian, 13 % of African, and 34 % of Oceanian children (Table 1). In the olderliterature from Europe and North America pneumococcus was reported in 32 %(81/255) and 10 % (63/645) of lung aspirates, respectively (Vuori-Holopainenand Peltola 2001). However, the results may have been influenced, especially inthe developing countries, by poor methodology (Scott and Hall 1999).

Some lung tap-based studies have related the pneumococcal etiology to thelobar and bronchopneumonia. S. pneumoniae was found in 78 % of patientswith lobar pneumonia, in contrast to only 13 % of those with bronchopneumonia(Vuori-Holopainen and Peltola 2001).

The incidence of pneumococcal disease is highest in young infants, with apredominance of males in all clinical manifestations (Klein 1981, Toikka et al.1999b). Among hospitalized children, about one-third have pneumonia and one-third a nonpulmonary infection such as meningitis or acute otitis media, the rest

22

having no identifiable focuses (Klein 1981). The peak incidence of hospitalizationfor pneumococcal pneumonia has been reported among infants from 13 to 18months of age (Klein 1981), but in a recent Finnish analysis of bacteremicpneumococcal pneumonia (N 85) the mean age was higher, 3.6 years (Toikka etal. 1999b).

5.3.1.4. Prevention of pneumococcal infection

The steady increase in pneumococcal resistance to several antimicrobials hasprompted prevention of pneumococcal infections by immunization.Polysaccharide vaccines consisting of 23 capsular types are available, but theyare poorly immunogenic in infants and young children (Mäkelä et al. 1983,Rubin 2000). Therefore, conjugate vaccines have recently been developed. Theyhave improved immunogenicity, and are more effective in eradicatingpneumococci from the nasopharynx (Rubin 2000). A heptavalent vaccine isestimated to be able to prevent up to 86 % of pneumococcal bacteremias and 83 %of pneumococcal meningitides in the USA (Butler et al. 1995). Recently, it hasbeen shown to reduce the number of pneumococcal otitis media episodes by 33 %(Eskola et al. 2001). The potential effectiveness against pneumonia is less wellknown, although recent preliminary results from the USA suggest a reductionof up to 35 % in the incidence of radiologically proven childhood pneumonia(Shinefield and Black 2000).

5.3.2. Haemophilus influenzae

Six encapsulated and several nonencapsulated (nontypable) strains of H.influenzae have been identified. The most pathogenic is type b, which causes 5 to18 % of cases of bacterial pneumonia in unvaccinated areas (Jadavji et al. 1997).The virulence of nontypable strains is lower, and asymptomatic children frequentlycarry these in their nasopharynges (Hjuler et al. 1995).

Nontypable H. influenzae is commonly found in patients with upperrespiratory infection or acute otitis media, and is probably also an importantcause of lower respiratory infections. Serologically verified nontypable H.influenzae infection is found in 6 to 17 % of children with community-acquiredpneumonia (Nohynek et al. 1991, Ruuskanen et al. 1992, Korppi et al. 1993a,Heiskanen-Kosma et al. 1998, Juvén et al. 2000). Up to 80 % of H. influenzaeinfections have been reported to occur as mixed infections with a virus or asdual infections with another bacterium (Korppi et al. 1993a). Convincingevidence for the pathogenicity of nontypable strains has been obtained frombronchoalveolar lavage (Wang et al. 1994) and lung aspirate cultures (Shann etal. 1984). According to lung tap studies, H. influenzae is found more often in

23

bronchopneumonia than in lobar pneumonia (20 % vs. 7 %) (Vuori-Holopainenand Peltola 2001).

The highest frequency of culture-positive H. influenzae pneumonia has beenreported from Papua New Guinea, where this agent was found in 80 % of lungaspirates, only 12 % being of type b (Shann et al. 1984). In contrast, in Africaand Asia, H. influenzae has been isolated only in 3 % and 1.4 % of lung aspirates,respectively (Table 1). These great differences may be explained by suboptimalmethods for detecting H. influenzae (Scott and Hall 1999).

5.3.3. Moraxella catarrhalisM. catarrhalis is an important cause of otitis media and sinusitis, but its role inlower respiratory infections is less clear. Serologic evidence for M. catarrhalishas been found in 2 to 10 % of hospitalized children with pneumonia (Nohyneket al. 1991, Ruuskanen et al. 1992, Korppi et al. 1993a, Heiskanen-Kosma etal. 1998, Juvén et al. 2000), but few studies have reported it from lung aspirates(Abdel-Khalik et al. 1938, Shann et al. 1984, Wall et al. 1986). Notably, M.catarrhalis often appears as a co-pathogen in both lung tap studies (Abdel-Khalik et al. 1938, Shann et al. 1984) and serology-based studies (Nohynek etal. 1991, Korppi et al. 1992a, Korppi et al. 1993a). This suggests that it israrely a primary cause of pneumonia in children.

5.3.4. Mycoplasma pneumoniae

M. pneumoniae and Chlamydia pneumoniae are naturally resistant to β-lactamantimicrobials. These bacteria cause epidemics at intervals of a few years(Hammerschlag 1995, Heiskanen-Kosma et al. 1999). Naso- and oropharyngealcarriage is common in asymptomatic children, especially during epidemic periods(McMillan et al. 1986, Foy 1993, Normann et al. 1998), and after symptomaticinfection, the carriage of M. pneumoniae may last for months (Foy 1993). Thus,detection of M. pneumoniae in the throat of a school-aged child does not implythat the current disease is caused by this organism (McMillan et al. 1986).

Mycoplasmal infections are common in children > 5 years of age, the incidencebeing highest at school age (Foy 1993). However, some recent data suggest itsimportance in younger children also (Hammerschlag 1995). Most often M.pneumoniae causes mild upper respiratory infections such as sore throat,pharyngitis, and tracheitis, with tracheobronchitis or pneumonia developing inonly 5 to 10 % of patients (Wendelien Dorigo-Zetsma et al. 2001). Probablybecause of its relative mildness, mycoplasmal pneumonia is found more frequentlyin ambulatory (20 to 40 %) than in hospitalized patients (10 to 20 %) (Foy

24

1993, Korppi et al. 1993a, Heiskanen-Kosma et al. 1998, Ruuskanen andMertsola 1999). Mycoplasma diagnostics are usually based either onmeasurement of serologic responses or on polymerase chain reaction (PCR)from naso- or oropharyngeal aspirates, but isolation of mycoplasma frompleural fluid provides convincing evidence for its pathogenic role in pediatricpatients, too (Loo et al. 1991).

5.3.5. Chlamydia pneumoniaeSerologic surveys have shown a rising prevalence of C. pneumoniae antibodieswith increasing age. Antibodies are found in 10 % of children from 5 to 10 yearsof age and in 30 to 45 % in adolescence (Grayston et al. 1990). However, mostinfections are mild or asymptomatic; clinical pneumonia occurs in only 10 %(Jadavji et al. 1997). In the USA, C. pneumoniae has been reported to beresponsible for 6 % of community-acquired cases of pneumonia in ambulatorychildren (Wubbel et al. 1999). In contrast, in a Finnish population-based studycomprising both ambulatory and hospitalized patients, C. pneumoniae was foundin 9 % of children between 5 and 9 years of age and in 31 % of those ≥ 10 yearsof age (Heiskanen-Kosma et al. 1999). In another Finnish series including onlyhospitalized children, C. pneumoniae was detected in 3 % of cases (Juvén et al.2000), suggesting that most C. pneumoniae infections are mild. C. trachomatismay also cause respiratory infections sometimes, especially in young infants inthe USA and in developing countries (Hammerschlag 1994).

5.3.6. Respiratory viruses

A wide range of viruses, such as RS, rhino-, parainfluenza, influenza, adeno-,corona-, entero-, varicella-zoster-, and human herpes virus type 6 (HHV-6),have been associated with childhood pneumonia (Jadavji et al. 1997, Ruuska-nen and Mertsola 1999). Overall, they have been estimated to account for 20 to62 % of cases, being found especially frequently in children < 2 years of age(Jadavji et al. 1997, Ruuskanen and Mertsola 1999, Wubbel et al. 1999, Juvénet al. 2000). RS virus is detected most often in small children, especially ininfants (Korppi et al. 1993a). It was found in 30 to 45 % of pneumonia casesduring an epidemic compared to 5 % in the non-epidemic season (Juvén et al.2000). Rhinoviruses are important causes of the common cold (Mäkelä et al.1998), but recently they have also been found in 24 % of nasopharyngeal aspirates(NPA) of children hospitalized for pneumonia (Juvén et al. 2000). Some further

25

evidence for the role of rhinoviruses in lower respiratory infections has beenobtained from studies on bronchial samples and post-mortem lung biopsies(Schmidt et al. 1991, Imakita et al. 2000).

In general, it is not known whether a virus isolated from the upper respiratorytract is correlated with infection in the lower respiratory tract, or has only pavedthe way for a bacterial pathogen. Hughes et al. compared virus isolations fromthe upper respiratory tract with those from lung aspirates. In two out of eightcases, the same virus (parainfluenza type 2, polio type 1) was found in bothsamples. In six cases, a virus was isolated only from the upper respiratory tract,but never only from the lung (Hughes et al. 1969). In a post-mortem study of13 lung tissue samples, viruses were isolated from the lung in six cases, three ofwhich were also found in the NPA (Avila et al. 1990). Viral diagnostics havebeen performed from lung aspirates obtained from consolidated pneumonia inat least nine studies (Vuori-Holopainen and Peltola 2001), the total identifi-cation rate being only 4 %. With the exception of a study from The Gambia, inwhich viruses were isolated in 30 % (28/94) of aspirates (Adegbola et al.1994), the virologic yield from lung aspirates has been very low. This lowyield may partly be explained by technical problems and lack of modernvirologic methods.

5.3.7. Mixed infections

Several, mostly serology-based, studies suggest that mixed viral-bacterialinfections are common, being found in 10 to 34 % of childhood pneumonias(Table 2). Bacterial coinfection has been detected with the greatest frequency inrespiratory syncytial virus (RSV) infections (Paisley et al. 1984, Turner et al.1987, Korppi et al. 1989, Korppi et al. 1993a). Recently evidence for concomitantbacterial infection was reported also in nearly half of patients with rhino-,parainfluenza -, or adenovirus associated pneumonia (Juvén et al. 2000).

Dual bacterial infections have also been detected in both serologic (Nohyneket al. 1991, Heiskanen-Kosma et al. 1998, Juvén et al. 2000) and lung aspiratestudies (Shann et al. 1984). A multiple bacterial etiology appears to be morecommon in children than in adults (Scott and Hall 1999). Besides being causedby two different species, dual infection may also occur within a single species.Two serotypes of pneumococci have been isolated simultaneously from lungand blood samples (Falade et al. 1997), and two different populations ofH. influenzae from lung aspirates (Shann et al. 1984).

26

5.4. Methods for identification of etiology

A wide range of microbiologic methods based on a variety of patient sampleshas been developed to identify the etiology of pneumonia and other invasiveinfections. In pneumonia, blood, serum, urine, throat swab, NPA, bronchoalveolarlavage, lung biopsy, pleural fluid and lung aspirate have been used. Involvementof a pathogen has been shown directly by culture, antigen detection, or nucleicacid detection, or by measuring specific antibodies or immune complexes.However, in obtaining valid information, several problems have emerged. Manymethods, especially in bacterial infections, have proven unreliable in terms ofeither low sensitivity or specificity, or the difficulty of obtaining representativesamples (Table 4). Cover of all potential pathogens requires a wide, expensive,and laborious panel of methods.

Table 4. Problems posed by different microbiologic methods in the search forbacterial etiology in pneumonia.

Method Problem

Culture

Blood Low sensitivity

Sputum Contamination with oropharyngeal flora

Oropharyngeal specimen High carriage rate of potential pathogenic bacteria

Bronchoalveolar lavage Requires anesthesia in children

Lung aspirate Invasive method, risk of pneumothorax

Other

Antigen detection Disputable sensitivity and specificity, depending on samples

Antibody assay Paired samples usually required, disputable sensitivity and specificity

Immune complex detection Disputable sensitivity and specificity

Nucleic acid detection Varying sensitivity, depending on samples

27

5.4.1. Bacterial agents

5.4.1.1. Culture

A positive culture from a normally sterile body site - in pneumonia from blood,pleural fluid, transtracheal aspirate, or lung aspirate – is considered the mostspecific indicator for microbiologic diagnosis of bacterial infection (Ruuskanenand Mertsola 1999). Studies of bacteremic diseases, however, indicate thatblood culture has a very low sensitivity (Isaacman et al. 1996). Negative results,especially in children, may reflect intermittent and low-density (≤ 10 to 15organisms per milliliter) bacteremia (Isaacman et al. 1996). Collecting a sing-le large-volume blood sample may slightly increase the sensitivity (Isaacmanet al. 1996).

Gram staining and culture of a well-produced sputum sample may bediagnostic in adults and in adolescents (Jadavji et al. 1997), but in children thesample is frequently contaminated with upper respiratory tract pathogens (Korppiet al. 1992b). For the same reason, naso- and oropharyngeal specimens cannotbe used (Klein 1981, Foy 1993, Normann et al. 1998).

Since, moreover, pleural fluid is seldom present in uncomplicated pneumonia(Pírez et al. 2001), invasive diagnostic techniques such as transtracheal aspiration,bronchoscopy with a protected brush catheter, bronchoalveolar lavage, and lungtap have been used. In clinical practice, however, their use has been restrictedby risk of potential adverse events.

5.4.1.2. Antigen detection

Soluble antigenic components of bacteria may be detected in a sample ofconcentrated urine, serum, sputum, pleural fluid, or lung aspirate (Farringtonand Rubenstein 1991, Boersma et al. 1991). Commercial agglutination reagentsare available for pneumococcus and Hib, but not for the other serotypes ofH. influenzae. Unfortunately, most assays have limited specificity. False-positiveresults may be obtained because of cross-reactions with other bacteria or mucosaldamage caused only by viruses, and serum or urine may be positive because ofpneumococcal infection at sites other than the lungs (Ramsey et al. 1986,Bromberg and Hammerschlag 1987, O’Neill et al. 1989, Korppi et al. 1993b).Furthermore, polysaccharide excretion may continue for several weeks afteracute infection or vaccination (Bromberg and Hammerschlag 1987, Kelley andKeyserling 1997). In respiratory secretions, it may be difficult to distinguishinfection from colonization (Korppi et al. 1992b). False-negative results from

28

serum samples are also relatively common (Silvermann et al. 1977, O’Neill etal. 1989). Consequently, despite the rapidity and technical simplicity of antigendetection assays, it may be doubted whether they are reliable diagnostic methodsin pneumonia.

5.4.1.3. Antibody and immune complex assays

Measurements of antibody responses and microbe-specific immune complexes(IC) from serum samples have been used in the etiologic diagnostics of lowerrespiratory infections in both adults (Lehtomäki et al. 1988, Jalonen et al. 1989,Lieberman et al. 1996) and children (Nohynek et al. 1995, Heiskanen-Kosma etal. 1998, Korppi and Leinonen 1997, 1998, Juvén et al. 2000).

A bacterial antibody response as such, may indicate various host-pathogenrelationships: carriage, carriage acquisition, subclinical infection, or clinicaldisease (Gwaltney et al. 1975, Nohynek et al. 1995). Furthermore, severalconcomitant clinical manifestations such as acute otitis media may cause thepositive serologic response (Leinonen et al. 1981, Virolainen et al. 1996, Rapo-la et al. 2001). On the other hand, viral infection does not often seem to causeseroresponse to bacterial agents. Only two of 200 young adults with commoncold developed positive antibody response to either S. pneumoniae or toH. influenzae (Mäkelä et al. 1998). Furthermore, in a study of 52 children withapparent viral laryngitis, only three had a positive antibody response to arespiratory bacterium (Korppi et al. 1990). In a series of 200 healthy adults andchildren, a diagnostic rise for S. pneumoniae, H. influenzae or M. catarrhaliswas detected in less than 1 % of paired sera (Korppi et al. 1989). The sensitivityand specificity of these assays in childhood pneumonia cannot be determinedprecisely, since serology has only rarely been compared with culture of bacteriaisolated from blood (Toikka et al. 1999a) and never with bacteria isolated froman infected lung.

5.4.1.3.1. Pneumococcal serology

Pneumococcal infection induces serum antibody responses to Ply, to C-PS, andto type-specific capsular polysaccharides, the first two being common to allpneumococcal strains. The enzyme immunoassay (EIA) for measuring IgGantibodies to Ply seems to be a useful tool for the diagnosis of pneumococcalpneumonia in both adults (Jalonen et al. 1989, Lieberman et al. 1996) and children(Korppi et al. 1992b, Nohynek et al. 1995), although the sensitivity of the assayis lower in children (Claesson et al. 1989). False positive results in children areconsidered rare; a diagnostic increase in titer was demonstrated in only 1 to 2.7 %of serum samples from healthy children (Korppi et al. 1993a, Nohynek et al.1995). The presence of otitis media or pneumococcal carriage was not associated

29

with any higher frequency of antibody response to pneumolysin in 121 childrenwith acute lower respiratory infection (Nohynek et al. 1995). In children withverified pneumococcal otitis media, IgA antibodies to Ply were usually found(Virolainen et al. 1996). Recently, however, it was shown that IgG antibodiesto Ply are also induced by carriage and by pneumococcal otitis media (Rapolaet al. 2001).

Measurement of type-specific capsular polysaccharides has proven sensitivein adults, but, in children < 2 years of age, antibody formation is weak (Claessonet al. 1989). In children, measurement of C-PS antibodies has also been used inthe serologic diagnosis of pneumococcal infections (Koskela 1987) includingpneumonia (Korppi et al. 1993b, Wubbel et al. 1999, Juvén et al. 2000). Falsepositive results in healthy children have been rare (Korppi et al. 1989).

In many invasive infections, including those caused by S. pneumoniae, antigensand antibodies are partly sequestered in ICs in the circulation (Korppi and Lei-nonen 1997, 1998). In assays measuring free antibodies and in antigen detectiontests, this may result in false-negative results. Methods for detectingpneumococcal ICs from either acute or convalescent serum samples have beendeveloped (Mellencamp et al. 1987, Leinonen et al. 1990, Holloway et al. 1993).Measurements of ICs specific to Ply and C-PS have been found to be moresensitive and specific than conventional antibody assays (Korppi and Leino-nen 1997, 1998). In infants, however, measurement of circulating ICs has beenas insensitive as other antibody assays (Korppi and Leinonen 1998). Furthermore,recently in a study comprising adults, Ply-ICs were shown to be present in acute-phase serum samples from healthy pneumococcal carriers more often than inthose from patients with bacteremic pneumococcal pneumonia (Musher et al.2001). In studies in which several serologic tests have been used for pneumococcaldiagnostics, often only one test has been positive (Korppi and Leinonen 1998,Heiskanen-Kosma 2000, Toikka et al. 2000). This suggests limited sensitivityof all the currently available tests.

5.4.1.3.2. Other bacterial serology

In contrast to pneumococcal strains, no tests for detecting antibodies againstantigens common to all H. influenzae or all M. catarrhalis strains have beencommonly available. Serum antibodies to these bacteria can be measured byEIA, with a mixture of whole cell antigens of 10 strains of H. influenzae orM. catarrhalis isolated from middle ear effusions (Burman et al. 1994).Significant rises in antibody titers were observed in only 2.7 and 0 % of healthychildren, respectively (Nohynek et al. 1995). Carriage of these bacteria may,however, induce some positive results. In 121 children with acute respiratoryinfection, positive antibody responses to H. influenzae or to M. catarrhalis

30

were found more often from children with simultaneous carriage of the bacteriumcompared to those with negative pharyngeal sample (Nohynek et al. 1995).

For Chlamydia, a complement fixation assay (CF) and EIA can be used as ascreening test, and species-specific diagnosis is possible with microimmuno-fluorescence (MIF) (Ekman et al. 1993). Carriage of chlamydia has rarely beenreported to cause a serologic response in children, which suggests a relativelyhigh specificity of serology in chlamydia diagnostics (Normann et al. 1998).However, the role of subacute or chronic infection as the cause of a seroresponseis not yet well known (Tuuminen et al. 2000).

For mycoplasma, the serologic methods most commonly used are CF andEIA. CF has been found to detect only 40 % of M. pneumoniae infections inchildren < 4 years of age (Kleemola et al. 1993). Measurement of IgM antibodiesby EIA is considered more sensitive and is more specific, even in young children(Waris et al. 1998).

5.4.1.4. Nucleic acid detection

A nucleic acid probe is a nucleic acid labeled with a compound that can bevisualized to detect a complementary sequence of DNA by hybridizing to it. Inthis way microorganisms can be detected with probes hybridizing with the genesof the organism. Too small a number of organisms is a limiting factor for theroutine use of gene probes and, therefore, PCR is increasingly used to amplifythe specific gene sequence. Thus, a very small number of pathogens can bedetected, which makes PCR not only sensitive but also prone to contaminationand false-positive results. On the other hand, natural inhibitors, e.g. porfphyrinfrom heme in clinical samples, may inhibit the polymerase enzyme (Higuchi et al.1989), resulting in false-negative results. Because PCR can detect fragments ofthe DNA of infectious agents, it is assumed that previous administration ofantimicrobials does not influence the PCR results as much as it does the cultureresults. However, no pneumococcal DNA was found in serum samples 48hafter initiation of intravenous ceftriaxone (Dagan et al. 1998). This is inaccordance with a Finnish study of pneumococcal bacteremia, in whichpneumolysin-PCR was negative in patients who had been on antimicrobials for2-5 days before the sample was taken (Toikka et al. 1999a).

For detection of pneumococcal bacteremia and pneumonia, Ply-PCR fromwhole blood, blood culture, serum, and buffy-coat samples has been used inboth adults (Rudolph et al. 1993, Salo et al. 1995, Lorente et al. 2000) andchildren (Dagan et al. 1998, Toikka et al. 1999a). The sensitivity of the assay

31

has varied considerably. In children with invasive pneumococcal infection orlobar or alveolar pneumonia, the pneumococcal genome was found in 15-43 %of blood or serum samples by Ply-PCR (Dagan et al. 1998, Toikka et al. 1999a).As compared with conventional serology and IC detection, blood-based Ply-PCR was a more sensitive method. However, its clinical value is reduced by thefact that, for optimal sensitivity, three blood fractions should be tested (Toikkaet al. 1999a). The specificity of the assay has been fairly good, thoughoropharyngeal strains (Dagan et al. 1998) and α-hemolytic streptococci, whenpresent in the sample (Kearns et al. 2000), may induce false-positive results. Inadults, another pneumococcal PCR, an assay to detect DNA of penicillin-bindingprotein, has also been used for lung aspirates; the PCR had a higher sensitivitythan the standard culture (Ruiz-Gonzalez et al. 1997).

PCR assays for H. influenzae, M. pneumoniae, and C. pneumoniae havebeen used in otitis media and pneumonia diagnostics for middle ear effusionsor NPA (Ueyama et al. 1995, Rayner et al. 1998, Jero et al. 1999, Waris et al.1998). In pneumonia, however, the relevance of a positive PCR result fromthe upper respiratory tract sample is questionable (Normann et al. 1998).With regard to the diagnosis of acute infections in general, results based ondetection of very small amounts of certain organisms may need to be interpretedwith caution.

5.4.2. Viral agents

In contrast to frequent nasopharyngeal carriage of bacteria potentially causinglower respiratory infections, carriage of viruses is less common. Therefore NPAsamples are more commonly used for viral diagnostics. Viral culture is the goldstandard (Bromberg and Hammerschlag 1987). More recently PCR assays forseveral common respiratory viruses have also been developed (Paton et al. 1992),though viral antigen detection from NPA samples is used more commonly(Bromberg and Hammerschlag 1987). In clinical practice viral serology is seldomneeded, but in a research setting it may be useful. CF and EIA are used mostoften, but the former has a relatively low sensitivity in childhood respiratoryinfections (Korppi et al. 1986, Avila et al. 1990). For optimal diagnostics ofrespiratory infections, the combined use of viral culture, direct antigen detection,and determination of IgG antibodies in paired sera is considered necessary(Korppi et al. 1993b).

32

5.4.3. Lung tap in childhood pneumonia

5.4.3.1. Earlier clinical experience

Lung tap, a method over a hundred years old (Leyden 1882) for obtaining arepresentative sample, provides several advantages as compared with the methodsdescribed above.1) The sample is obtained straight from the infectious focus, the risk of

contamination with normal flora thus being avoided (Sinha and Huges1966, Davidson et al. 1976).

2) Some organisms rarely cause bacteremic infections, but may be isolatedfrom the lung. In a study on Kenyan adults, H. influenzae was isolatedfrom 10 of 259 lung aspirates but from only 1 of 518 blood cultures(Scott et al. 2000).

3) Culture of a lung aspirate has a very low false-positive rate; a positiveculture is considered strong evidence for bacterial infection (Shann 1986).In Chile, a lung tap was performed on 13 infants with no clinical orradiographic evidence of pneumonia; all 13 cultures remained negative(Mimica et al. 1971).

According to data on more than 3500 children on whom lung tap was performed,the method showed a relatively high positivity rate for bacterial agents (Shann1986, Scott and Hall 1999, Vuori-Holopainen and Peltola 2001). A bacterialetiology was revealed in 50 % of children with pneumonia overall (Vuori-Holo-painen and Peltola 2001), the positive bacterial yield ranging from 10 to 96 %(Vuori-Holopainen and Peltola 2001). Several reasons have been suggested forfalse-negative results: the needle may have missed the consolidation, too smalla specimen or none may have been obtained, previous antimicrobial therapymay have disturbed the growth, or the organism may not have grown in themedia used, or there has been a delay in culturing the specimens (Shann 1986,Scott and Hall 1999). The positivity rate of the tap improves with the size of theconsolidation, dense peripheral consolidations being optimal (Zalacain et al.1995). Other factors associated with higher sensitivity are the experience of theperson performing the tap, and lack of previous antimicrobial therapy (Zalacainet al. 1995). In contrast to bacteria, few studies have explored viruses in lungaspirates; therefore no conclusions can be drawn concerning the applicability oflung tap in viral diagnostics (Vuori-Holopainen and Peltola 2001).

5.4.3.2. Risks of the procedure

Because lung tap is an invasive method, it is potentially harmful, the main riskof the procedure being iatrogenic pneumothorax. In a review of 2 658 childrenon whom lung tap was performed, 0.5 % required pleural drainage (Vuori-

33

Holopainen and Peltola 2001). The risk of clinically significant pneumothoraxmay be reduced by puncturing the lung once only, by restricting the use of themethod to experienced operators, and by using a thin needle (Scott and Hall1999). Other, even rarer, adverse events described in children are hemoptysis(1.3 %) and subcutaneous emphysema (0.2 %) (Vuori-Holopainen and Peltola2001). Pleuritic pain has rarely been reported. Three deaths associated withlung tap have been described in children, in two of which the lung tap in itselfmay have been the primary cause of death (Wesley et al. 1971, Cunanan et al.1979, Shann et al. 1984). Notably, most of the lung-tap studies have includedonly patients with consolidated pneumonia (Vuori-Holopainen and Peltola 2001).Thus, in other types of pneumonia, the applicability and risks of the method aredifficult to assess.

5.5. Acute host response parametersdistinguishing viral from bacterial etiology

Erythrocyte sedimentation rate (ESR), peripheral white blood cell count (WBC),and CRP are common host-response parameters used to differentiate bacterialfrom viral infection. For ESR, a cut-off value ≥ 30 mm/H has been related to anincreased risk of bacteremia or radiologically verified pneumonia (McCarthy etal. 1977). However, the response of ESR is rather slow (Triga et al. 1998) andmay be influenced by the size and shape of the erythrocytes, and the presence ofimmunoglobulins (Gabay and Kushner 1999).

The WBC response is much more rapid than that of ESR. In children withlobar or segmental pneumonia, the WBC count is at its highest during the firsttwo days of illness, and lowest on the fourth day (Triga et al. 1998). The mostcommon cut-off level used for differentiating between viral and bacterial etiologyis ≥ 15 x 109 /L (McCarthy et al. 1977, Putto et al. 1986). However, controversyexists as to whether any level of WBC distinguishes a bacterial from a viralrespiratory infection reliably (Korppi et al. 1993c), because leukocytosis mayalso occur in viral infection, especially if caused by adenovirus or Epstein-Barrvirus (Putto-Laurila et al. 1999). In a series of 154 febrile children, a WBCcount ≥ 15 x 109 /L was found in 67 % of cases with evidence for bacterialetiology vs. in 13 % of those with probable viral disease (Putto et al. 1986).

CRP is an acute phase protein synthesized within 6-8 hours by the liver inresponse to infection and a variety of inflammatory stimuli such as immunological,traumatic, ischemic, and malignant tissue damage (Du Clos 2000). It is especiallyuseful for monitoring the presence of focal infections such as septic arthritis andosteomyelitis, and their successful treatment (Unkila-Kallio et al. 1994, Kallio

34

et al. 1997). In childhood meningitis and bacteremic infections, serum CRP is asensitive and rapidly responding parameter differentiating bacterial from viraletiology (Peltola 1982, Peltola and Räsänen 1982, Peltola and Jaakkola 1988,Sormunen et al. 1999).

In respiratory infections, the ability of CRP to distinguish the etiology is lessevident, since CRP values of up to > 100 mg/L have been observed in infectionscaused by adenovirus, influenza virus, or RS virus (Ruuskanen et al. 1985). Insome cases, however, undetected concomitant bacterial infection may have beenpresent.

Nevertheless, high CRP values are mostly associated with bacterial disease. Ifa child has been ill for more than 8 to 12 hours, and the CRP value is < 40 mg/L,a viral etiology is highly probable (Peltola 1983, Putto et al. 1986). Recently,CRP was studied in children with bacteremic pneumococcal pneumonia. InitialCRP values of ≤ 20 mg/L, 20 - 59 mg/L, 60 - 179 mg/L, and ≥ 180 mg/L wereobserved in 15 %, 17 %, 42 %, and 26 % of patients, respectively (Toikka et al.1999b). The CRP value increased for 24 hours in 74 % of patients duringantimicrobial treatment, but declined rapidly after the first day of therapy (Toikkaet al. 1999b).

More recently the use of other host response parameters such as procalcitoninhas been studied (Genrel and Bohuon 2000). In general, routine laboratorytests have considerable, although limited, use in differentiating bacterial fromviral etiology, but they are of little value in specific identification of the causativepathogen.

5.6. Antimicrobial therapy

The use of antimicrobials has steadily increased worldwide. Respiratory infectionsaccount for 75 % of all community prescriptions, with approximately 818 millionprescriptions each year (Carbon and Bax 1998). Consequently, the antimicrobialsensitivities of several bacteria have decreased markedly during recent years.The spread of resistance reflects not only the appropriate use of antimicrobialsbut also inappropriate prescribing. Antimicrobials are frequently prescribed fornon-bacterial infections, broad-spectrum regimens are overused, and dosagesthat are too small or treatments that are too long are commonly prescribed(Arason et al. 1996, Guillemot et al. 1998, Amsden 1999).

35

5.6.1. Treatment in relation to antimicrobial resistance

The in vitro resistance of bacteria is defined in terms of the minimal inhibitoryconcentration (MIC), the lowest concentration that inhibits the growth of theorganism. For effective treatment, peak tissue levels of an antimicrobial shouldbe four to eight times the MIC values, and the concentration of the antimicrobialshould be higher than the MIC value for most of the time between doses (Shannet al. 1987). However, standard susceptibility testing methods fail to take intoaccount the bactericidal activity of human serum and the potential for phagocytesto enhance the antibacterial effects of antimicrobials intracellularly, both of whichmay restrict the spread of infection. Therefore, in vitro resistance defined byMIC values is not synonymous with failure of clinical treatment.

5.6.1.1. Streptococcus pneumoniae

Pneumococci are not intrinsically resistant to penicillin or to other commonlyused antimicrobials, but acquired resistance is increasing. The extent of theresistance depends on the geographical area, the age of the patient population,and the specimens investigated (Klugman 1990, Thornsberry et al. 1999, Forward1999). S. pneumoniae isolates with penicillin MIC < 0.1 µg/mL are defined aspenicillin-sensitive, strains with MIC 0.1-1.9 µg/mL as intermediate resistant,and strains with MIC ≥ 2 µg/mL as highly resistant to penicillin (Friedland andMcCracken 1994). Approximately one-third of S. pneumoniae isolates in theUSA shows some resistance to penicillin. In Europe, the proportion varies from1 to 50 % (Arason et al. 1996, Baquero 1995, Pradier et al. 1997), being highestin Spain and lowest (1-3 %) in Scandinavia and in the United Kingdom (Pradieret al. 1997). In Finland, the prevalence of highly resistant pneumococci is only1 % (Pradier et al. 1997, Finres 1999).

There is little evidence that infections caused by penicillin-resistantpneumococcus outside the central nervous system respond poorly to parenteralpenicillin or other β-lactams (Friedland and McCracken 1994, Friedland 1995,Pallares et al. 1995). In pneumonia, the concentration of penicillin in the lung isequal to the concentration in the serum (Shann et al. 1987). With parenteralpenicillin, peak serum concentrations of up to 15-18 µg/mL and concentration≥ 1 µg/mL for at least 18 hours are achieved (Shann et al. 1987). Hence, theconcentration achieved with parenteral penicillin is sufficient for treatment ofmost pneumococcal infections.

A potential problem with the use of penicillin, however, is that it does noteffectively eradicate the carriage of penicillin-resistant pneumococci. Thus, at

36

least in theory, there is a risk of selection for carriage of these strains, and anincrease of invasive diseases due to these resistant strains (Klugman 1990).This may be avoided by using parenteral therapy, with which higher antimicrobialconcentrations are obtained in nasopharyngeal mucosa (Klugman 1996).

Parallel with the spread of penicillin resistance, there has been an increase inthe resistance to other antimicrobials such as macrolides (Thornsberry et al.1999, Forward 1999). In the USA, up to 23 - 31 % of pneumococcal isolatesare resistant to erythromycin (Thornsberry et al. 1999, Gay et al. 2000). InFinland, the corresponding level is 7 % (Finres 1999), but, in the Helsinki area,12 % (Dr M. Vaara, personal communication 2001). The clinical relevance ofhigh-level macrolide resistance suggests that potential treatment failures mayincrease (Gay et al. 2000).

5.6.1.2. Influence of βββββ-lactamase production

Many H. influenzae and M. catarrhalis strains produce β-lactamase, whichcompromises the activity of penicillin, ampicillin, and amoxicillin, but doesnot substantially affect the activities of cephalosporins (Zhanel et al. 2000).The prevalence of β-lactamase production has hardly changed in the last fewyears (Thornsberry et al. 1999, Zhanel et al. 2000). In Canada and the UnitedKingdom, up to 48.7 % and 15.1 % of H. influenzae and 94 % and 93 % ofM. catarrhalis respiratory tract isolates, respectively, produce β-lactamase(Felmingham et al. 1998, Doern et al. 1999, Zhanel et al. 2000). In the Helsinkiarea, 20 % of all H. influenzae isolates and 93 % of the M. catarrhalis isolatesare β-lactamase-positive (Dr M Vaara, personal communication 2001).

Although failures of amoxicillin treatment in otitis media caused by β-lactamase-producing H. influenzae and M. catarrhalis have been described(Rosenfeld et al. 1994, Brook and Yocum 1995), β-lactamase production hasnot been shown to have as much influence on the management of pneumonia inimmunocompetent patients (Klugman 1996). In fact, in a Danish study, penicillinand amoxicillin were effective in vivo in eradicating β-lactamase-positiveM. catarrhalis from children with lower respiratory infections (Ejlertsen andSkov 1996). Since pulmonary infections caused by M. catarrhalis are rare inchildren, M. catarrhalis etiology need not to be considered in selection ofantimicrobials in cases of middle or lower respiratory infections affecting primarilyhealthy children (Korppi et al. 1992a).

5.6.2. Antimicrobial treatment studies

A number of studies have compared two antimicrobial regimens in lowerrespiratory infections. In adults, the efficacy of two β-lactams, cefuroxime and

37

ampicillin (Pines et al. 1981), and a β-lactam vs. a macrolide have been compared(MacFarlane et al. 1983, MacFarlane et al. 1996). Cefuroxime led to asatisfactory clinical response more often than ampicillin (Pines et al. 1981). Incontrast, a macrolide provided no advantage over standard amoxicillin(MacFarlane et al. 1996).

Treatment of childhood pneumonia has been studied, especially in developingcountries. In those countries, co-trimoxazole, a treatment widely used in pediatricpneumonia in the developing world, has been compared with procaine penicillinand ampicillin (Campbell et al. 1988, Keeley et al. 1990, Sidal et al. 1994,Straus et al. 1998). It was found effective in outpatient treatment. However, inmore severe pneumonia, procaine penicillin and ampicillin were more likely tobe effective. In probable pneumococcal pneumonia in Brazilian children, nodifference was found between a single dose of long-acting benzathine penicillinand a 7-day course of procaine penicillin (Camargos et al. 1997).

From the developed countries, only limited data are available from randomizedstudies. Atzithromycin was as effective as amoxicillin-clavulanate or erythromycinfor pediatric pneumonia (Roord et al. 1996, Harris et al. 1998). Intramuscularpenicillin and oral amoxicillin were found to be similar in effectiveness in theearly outpatient treatment of pediatric patients with presumed bacterialpneumonia (Tsarouhas et al. 1998).

5.6.3. Length of antimicrobial treatment

The length of treatment has also been mostly empirical and rarely based oncomparative studies. Shorter courses of antimicrobials have only recently beenvalidated in various invasive and focal infections such as acute cystitis of women,meningitis and osteomyelitis (MacFarlane et al. 1979, Lin et al. 1985, Peltola etal. 1989, Norrby 1990, Peltola et al. 1997). In childhood upper respiratoryinfections such as otitis media, sinusitis, and tonsillopharyngitis, short coursesare also effective (Pichichero and Cohen 1997, Cohen et al. 2000). In the latter,6-day therapy with amoxicillin or 4-5-day therapy with various cephalosporinsor with azithromycin have been suggested as alternatives to traditional 10-daypenicillin therapy (Pichichero and Cohen 1997). A recently published meta-analysis of randomized, controlled trials for the treatment of otitis media suggeststhat a 5-day course of short-acting antimicrobial is appropriate (Kozyrkyj et al.1998). Apart from studies showing that a 3- or 5-day course of azithromycin isas effective as a 10-day course of erythromycin or amoxicillin-clavulanate inpediatric pneumonia (Roord et al. 1996, Harris et al. 1998), there have beenvirtually no data on the length of treatment in childhood lower respiratory andnonfocal infections.

38

5.6.4. Current guidelines for treatment of pneumonia

Because of the lack of randomized controlled trials, most treatment guidelinesare based on observations of the organism’s in vitro susceptibility rather than onproof of benefit from one antimicrobial over another (Jadavji et al. 1997).However, in the absence of an etiologic diagnosis, the treatment of childhoodpneumonia is almost always empirical. β-lactams have long been the first choicefor treating childhood pneumonia, but recently, in accordance with guidelinesfor adulthood pneumonia (Am Thor Soc 1993), macrolides are often consideredthe first choice for ambulatory patients (Ruuskanen and Mertsola 1999). Noclear consensus exists, since many other experts still favor amoxicillin as theinitial treatment (Klugman 1996). In Scandinavian countries, penicillin G is thefirst choice in hospitalized children (Ruuskanen and Mertsola 1999). For patientsrequiring intensive care, a combination of cefuroxime and a macrolide is oftenrecommended (Jadavdi et al. 1997).

6 AIMS OF THE STUDY

6.1. General aim

The aim of this study was to investigate the etiologic diagnostics and treatmentof common invasive bacterial infections of children, with special emphasis onalveolar pneumonia.

6.2. Specific aims

1) To study the etiology of childhood community-acquired pneumonia andother invasive suspected bacterial infections applying serologic methods.

2) To compare the clinical efficacies of parenteral penicillin and cefuroximefor the treatment of common childhood infections requiringhospitalization.

3) To compare the clinical efficacies of a 4-day treatment and a 7-daytreatment, using these β-lactams.

4) To explore the advantages and disadvantages of lung tap in the etiologicdiagnostics of hospitalized childhood consolidated pneumonia.

39

7 PATIENTS AND METHODS

Table 5. Patients included in the study.

Paper I 170 children with suspected invasive bacterial infection

Paper II 154 children from paper I treated with randomized parenteral antimicrobials

(procaine penicillin N 75, cefuroxime N 79)

Paper III 154 children from paper I treated with parenteral antimicrobials,

the length of treatment randomized (4 days, N 73; 7 days, N 81)

Paper IV One child with pneumonia caused by Moraxella osloensis

Paper V 34 children with community-acquired consolidated pneumonia

7.1. SE-TU study (I-III)

7.1.1. Study design

The SE-TU (“Sepsis-tutkimus”, “Sepsis-research”) study was an open,randomized, prospective study of the etiology and treatment of commonchildhood infections, in which two antimicrobial regimens and two lengths oftreatment were compared. The study was carried out during two time periods(February 1988-October 1989, and March 1991-October 1993) at the Children’sHospital, University of Helsinki, and Aurora Hospital, Helsinki. The EthicsCommittees of both participating hospitals approved the SE-TU study protocol,and oral consent was obtained from the guardian of each patient. From acomputer-based list, one of the following four treatments was chosen: 1) procainepenicillin 50,000 IU/kg/day i.m. for 4 days, 2) procaine penicillin 50,000 IU/kg/day i.m. for 7 days, 3) cefuroxime 100 mg/kg/day i.v. divided into 3 doses for 4days, or 4) cefuroxime 100 mg/kg/day i.v. divided into 3 doses for 7 days. If thechild was known to be allergic to the randomized drug, the other alternativewas chosen. If clinical recovery was rapid, the 7-day treatments were completedat home with oral penicillin V or cefuroxime axetil, respectively.

40

If the child showed symptoms and signs suggesting an invasive bacterialinfection (McCarthy et al. 1982), he or she was included regardless of theresults of laboratory or radiographic investigations. The exclusion criteria were:age < 3 months, meningitis, osteoarticular, or urinary tract infections, patientsin the oncology, trauma, surgery, and intensive care units. CRP was measuredon admission (Peltola et al. 1984). If the value was ≥ 80 mg/L, the randomizedantimicrobial was instituted; if it was 20-79 mg/L, the measurement was repeatedwithin 8-12 hours, and the child was included if there was an increase of at least50 %. If CRP remained < 20 mg/L, antimicrobial was begun only if the patientdeteriorated. Blood cultures, NPA samples, and acute- and convalescent-phaseserum samples were collected. The etiology of infections was studied with anextensive panel of antigen and antibody assays detecting seven bacteria, 15 viruses,and one protozoan. The patients included in the SE-TU study and randomized forparenteral antimicrobials are presented in the study profile in Figure 1.

Originally, 178 patients fulfilled the inclusion criteria. The final number ofpatients in the etiologic analysis (Paper I) was 170. Ninety-seven children (57 %)were boys and 73 (43 %) were girls. Their ages ranged from 3 months to 15years, the mean age was 3.8 years (Figure 2).

The treatment analyses (Papers II and III) comprised 154 children. The reasonsfor exclusion of eight children enrolled in the etiologic analysis (Paper I) and of24 patients in treatment analyses (Papers II, III) are described in Figure 1.

Figure 1. Study profile of the SE-TU studies (I-III).

The SE-TU Study Profile

One case of likely viral disease, two patients refused to stay in hospital, and treatment was continued with another antimicrobial,two cases of prolonged treatment for otitis media or streptococcal tonsillitis, once insurmountable problems with i.v. access.

**

Four cases of likely viral disease, four prolonged treatment for otitis media or streptococcal tonsillitis, one case of probable listeriosis,one with change of treatment for undetermined reasons.

*

Patients with underlyingdisease excluded

N 8

Cefuroxime N 854-day N 397-day N 46

Patients with randomizedtreatment

N 170

Patients enrolledN 178

Penicillin N 854-day N 427-day N 43

10 patientsexcluded*

Study completedPen 4-day

N 36

Study completedPen 7-day

N 39

Study completedCfm 4-day

N 37

Study completedCfm 4-day

N 42

6 patients excluded**

41

Figure 2. Age distribution of 170 children in the SE-TU series (I).

7.2. KE-TU study (IV, V)

7.2.1. Study design

The KE-TU (“Keuhkokuume-tutkimus”, “Pneumonia-research”) study wasanother prospective study on the use of lung tap in pediatric pneumonia carriedout at the Helsinki University Central Hospital, Hospital for Children andAdolescents from December 1997 to December 2000. The Ethics Committeeof the hospital approved the study, and written consent was required from theparents. The diagnosis of pneumonia was based on simultaneous findings of atleast a history of fever and/or respiratory symptoms and consolidated infiltratescompatible with alveolar pneumonia in the chest radiograph. In all, 47 patientswere asked to participate, but, since 13 refused, the study included 34 hospitalizedchildren, in all of whom a lung tap was performed. Postoperative, oncologic,and immunocompromised patients, and those with a bleeding diathesis or inintensive care were excluded, as were those < 3 months of age. In paper IV, a6-year-old girl with bilateral pneumonia caused by Moraxella osloensis isdescribed. Paper V comprises 34 children with consolidated pneumonia. Themean age was 5.1 years, their ages ranging from 10 months to 14.0 years. Twenty(58 %) of the patients were boys and 14 (42 %) were girls.

42

7.2.2. Technique of lung tap

The consolidated lesion was localized in two planes according to chestradiographs and physical signs. The lung tap was performed in the outpatientdepartment, at the bedside without fluoroscopy or computed tomography, andbefore antimicrobial therapy was instituted. The skin was anesthetized withlidocaine (Emla®) ointment. During the procedure, the child was held tightly ina sitting or supine position. After sterilization of the skin, a standard thin needle(22G) was attached to a 5-mL syringe containing 2.5 mL of saline, and insertedsupracostally into the consolidated lung. To increase the volume of the aspirate,1.5 mL of saline was injected into the lung, and the needle was removedimmediately during continuous suction. The whole procedure lasted < 10seconds. After the lung tap, the patient was monitored closely and the respiratoryrate was checked repeatedly. A control chest radiograph was taken within 2hours after the procedure (Perlmutt et al. 1986).

7.3. Specimens and microbiologic methods

7.3.1. Conventional microbiology (I-V)

Blood cultures were obtained from all patients by peripheral venipuncture. Elevenpercent of the patients in the SE-TU series and 15 % of patients in the KE-TUstudy had received antimicrobials prior to hospitalization. For culture, the routine(BacT/Alert®) aerobic blood culture medium was used. The oropharyngealsamples obtained on admission were cultured routinely. In the SE-TU series (I-III), these samples were available only from patients with tonsillitis, but, in theKE-TU series (IV, V), from all patients. For viral antigen detection indirectimmunofluorescence from NPA samples was performed, using seven monoclonals(Respiratory Panel 1, Light Diagnostics, Temecula, CA, USA) against thefollowing viruses: RS, adeno, parainfluenza 1-3, and influenza A and B (Ukko-nen and Julkunen 1987). NPA samples were collected from 122/170 patients ofthe SE-TU series (I) and from 32/34 patients of the KE-TU series (V).

7.3.2. Serology (I-III)

For serology, acute-phase serum samples were collected on admission to hospitaland the convalescent sera 2-3 weeks later. All samples were stored at –70°Cuntil assayed. Microbiologic diagnoses were based on a single high value (IgM),a rise in the specific antibody titer in paired sera, or demonstration of specificICs in acute or convalescent serum (Table 6). The antibody assays for S.

43

pneumoniae, H. influenzae, and M. catarrhalis used in this study have beenvalidated in healthy children and in young adults with common colds; positiveresults were found ≤ 2.7 % of cases (Korppi et al. 1989, Nohynek et al. 1995,Korppi and Leinonen 1998, Mäkelä et al. 1998). Samples for determination of

Agent Antigen used Assay Diagnostic criteria Reference

S. pneumoniae Ply EIA 2-fold increase Jalonen et al. 1989

ICs Ply EIA GM+2SD* (≥550) Leinonen et al.1990,Korppi andLeinonen 1998

ICs C-PS EIA GM+2SD (≥400) Leinonen et al.1990,Korppi andLeinonen 1998

H. influenzae, whole cell EIA 3-fold increase Burman et al.1994nontypable

M. catarrhalis whole cell EIA 3-fold increase Burman et al.1994

Chlamydiae common group CF 4-fold increase Ukkonen et al. 1984antigen

C. pneumoniae type specific micro-IF 4-fold increase/ Ekman et al. 1993elementary bodies presence of IgM

C. trachomatis type specific micro-IF 4-fold increase/ Ekman et al. 1993elementary bodies presence of IgM

M. pneumoniae CF 4-fold increase Ukkonen et al. 1984whole cell EIA 3-fold increase Kleemola et al. 1990

S. aureus T5, T8 CPS EIA 2-fold increase, Fattom et al.1990,1993GM+2SD (≥25)

ICs alphatoxin EIA GM+2SD (≥500) Julander et al.1983

Respiratory and CF 4-fold increase Ukkonen et al. 1984other viruses**

Epstein-Barr virus IF presence of IgM Hedman et al.1993EIA IgG avidity <25% Hedman et al.1993

Human herpes virus 6 IF 4-fold increase Linnavuori et al. 1992

Toxoplasma gondii CF 4-fold increase Ukkonen et al. 1984

EIA IgG avidity <15% Hedman et al. 1993

* GM+2SD, geometric mean + 2 standard deviations

** Viruses included in the CF panel: influenzae A and B, parainfluenzae 1, 2, and 3, adenovirus, RSV, coxsackie B5,herpes simplex, rotavirus, cytomegalovirus, varicella zoster and mumps virus

Table 6. Serologic methods used (I-III).

44

S. pneumoniae, H. influenzae, M. catarrhalis and S. aureus antibodies andpneumococcal and staphylococcal ICs were available from all (N 170) patients,and for chlamydia, mycoplasma, EBV, and HHV-6 serology from 169, 127,165, and 161 patients, respectively. A CF assay for viruses and detection ofToxoplasma gondii antibodies was possible from 105 patients.

7.3.3. Lung aspirates (IV, V)The presence of leukocytes in the stained specimen was recorded; any leukocytesseen in the lung aspirate samples at 100x magnification were considered indicativeof a representative sample (Zalacain et al. 1995). Besides conventional stainingand culture methods, PCR techniques and commercial RNA-DNA hybridizationassays were also used to detect bacteria. The specific bacteria searched for withPCR were S. pneumoniae, H. influenzae, C. pneumoniae and M. pneumoniae(Table 7). The Gram-negative bacillus isolated from a cultured needle (paperIV) was further identified by DNA sequencing (Jalava et al. 1995, Jalava andEerola 1999).

Virus detection from lung aspirates was performed by means of viral culture,antigen detection (Respiratory Panel 1, Light Diagnostics, Temecula, CA, USA)and PCR detecting RNA specific for entero- and rhinoviruses (Lönnrot et al.1999). A summary of the microbiologic methods used and the number childrenon which each test was performed are presented in Table 7. If the amount of thelung aspirate sample was insufficient, priority was given to standard microbiologicstudies and aspirate was frozen (-70°C) for PCR studies only after taking samplesfor the staining and culture.

7.3.4. Serotyping and susceptibility testing ofStreptococcus pneumoniae (V)

The antimicrobial sensitivity of the S. pneumoniae strains was measured by thestandard disk diffusion method. If the test indicated lowered antimicrobialsusceptibility, MIC-concentrations were determined by the E test (Forward 1999).Pneumococcal isolates were serotyped by counter-immuno-electrophoresis andby the latex agglutination test for neutral polysaccharides of types 7 and 14,utilizing omniserum, serum pools, and type-, group-, and factor-specific antisera(Statens Serum Institut, Copenhagen, Denmark). In unclear cases, the capsularquellung reaction was used for confirmation of the result.

45

7.4. Criteria for etiology (I-V)

Papers I-III. An infection was considered to be of bacterial origin if at least oneof the following criteria was fulfilled: positive blood culture, a significant antibodyresponse or specific ICs present in sera samples (Table 6), or Streptococcuspyogenes isolated from the pharyngeal sample. The diagnosis of viral infectionwas based on virus antigen detection from NPA or a significant antibody response(Table 6). A mixed viral-bacterial infection was diagnosed when there wasmicrobiologic evidence for bacterial and viral infection in the same patient. Dual

Pathogen Method No of tests Referenceperfromed

Bacteria Staining (acridine orange/gram) 34/34 Baron et al.(ed.) 1994,Culture (Bact/Alert®) 34/34 Lauer et al.1981Needle culture 33/34

S. pneumoniae Pneumolysin PCR 32/34 Kontiokari et al. 2000

DNA-RNA hybridization 29/34(Accuprobe®)

C. pneumoniae PCR 32/34 Described in paper V

H. influenzae PCR 32/34 Hobson et al. 1994

DNA-RNA hybridization 27/34(Accuprobe®)

M. pneumoniae PCR 28/34 van Kuppeveld et al. 1992Vesanen et al. 1996

M. osloensis DNA sequencing 1/34 Jalava et al.1994, 1995

Viruses Culture 33/34

Adenovirus IF (Light Diagnostics) 26/34Influenza A and B IF (Light Diagnostics) 26/34Parainfluenzae 1-3 IF (Light Diagnostics) 26/34RSV IF (Light Diagnostics) 26/34

Enterovirus PCR 28/34 Lönnrot et al. 1999Rhinovirus PCR 28/34 Lönnrot et al. 1999

Table 7. Microbiologic methods used for lung aspirates (IV, V).

46

and triple infections were considered to be present when evidence for two orthree bacteria or viruses was found in the same patient. The percentages werecounted from all patients, although some assays could not be performed fromthe whole series.

Papers IV and V. The etiology of pneumonia was considered definitivewhen a pathogen was detected in a normally sterile sample (blood, lung aspirate,or puncture needle). The etiology was considered presumptive if a viral antigenwas detected in the NPA, or if S. pyogenes was isolated from the pharyngealsample of a patient with tonsillitis.

7.5. Radiologic assessment (I-V)

Papers I-III. Chest radiographs were obtained on admission and, if the findingsuggested pneumonia, a control radiograph was taken five days later. Theradiographs were reviewed by a pediatric radiologist unaware of the results ofmicrobiologic assays or clinical signs. Patients with alveolar consolidation, eitherlobar or bronchopneumonia, were included in the pneumonia group, whereasthose with respiratory symptoms but no clear alveolar consolidation, such asdiffuse and increased interstitial infiltrates, were included in the group of “otherrespiratory infections”.

Papers IV and V. The diagnostic criteria for pneumonia were a consolidatedalveolar infiltration compatible with pneumonia observed on admission, togetherwith fever and/or respiratory symptoms. Radiologically pneumonia was definedas a consolidated infiltrate, which diminished or disappeared during the follow-up. Chest radiographs were interpreted by the radiologist on duty, and re-evaluated by another pediatric radiologist unaware of the etiologic diagnosisand previous radiologic results. Check-up radiographs were similarly interpreted.

7.6. Statistical methods (II, III, V)

In the treatment analyses, the primary endpoint for successful treatment wasuneventful recovery with no change in the randomized antimicrobial (II) or inthe length of treatment (III). A sample size analysis showed that a total of 142patients would have disclosed a possible 20 % difference in the effectiveness ofthe two treatments at a level of p = 0.05 (two-sided test) with an 80 % power(II, III). For categorical comparisons, the chi-squared contingency analysis andFisher’s exact test were used. For continuous variables, analysis of variance andStudent’s t tests were used to compare the mean values (II, III, V). All testswere two-sided. P < 0.05 was considered statistically significant.

47

8 RESULTS

8.1. Etiology in the four clinical manifestations (I)

The patients in this analysis were divided into the following four categories:1) Pneumonia (N 79). Children with acute respiratory symptoms and signs

and alveolar consolidation in a chest radiograph.2) Other respiratory infections (N 30). Patients with respiratory symptoms

and signs, but without a clear radiologic finding suggesting alveolarpneumonia.

3) Sepsis-like infections (N 37). Children with high fever (≥ 38.5°C), and aninitial CRP value ≥ 100 mg/L but no respiratory symptoms or evidencefor a focal infection.

4) Other acute infections (N 24). Cases with no respiratory symptoms, highfever, or clearly increased CRP (≥ 100 mg/L) value but in which theclinical status warranted antimicrobials (e.g. cellulitis, lymphadenitis,sepsis-like disease with either lower fever or CRP < 100 mg/L).

Evidence for a microbial etiology was found in 61 %, 73 %, 68 %, and 42 % ofthe groups of pneumonia, other respiratory, sepsis-like, and other acute infections,respectively (Table 8). Bacteria alone were found in 49 %, 50 %, 46 %, and 38 %of children with pneumonia, other respiratory, sepsis-like, and other acuteinfections, respectively, while mixed bacterial-viral infections were found in 11 %,7 %, 3 %, and 0 % of cases, respectively. In all groups combined, bacterial etiologyalone, or concomitantly with viruses was found in 92 children. The overalldistribution of bacterial, mixed, and viral etiologies in the various clinicalmanifestations in patients with etiology identified is shown in Figure 3.

S. pneumoniae was the most common bacterial pathogen found (N 62), beingpresent in 46 %, 30 %, 24 % , and 33 % of pneumonia, other respiratoryinfections, sepsis-like infections, and other acute infections, respectively. Eitheralone or combined with other agents, it accounted for 75 %, 41 %, 36 %, and80 % of cases with an identified etiology. The next commonest bacterial pathogensidentified were nontypable H. influenzae and C. trachomatis (Table 8). Mostbacterial diseases were diagnosed by serology alone, since blood culture waspositive in only 5 %.

48

Disease Etiology identified Agent N %N % (in each category)

Pneumonia 48 61 S. pneumoniae alone 24 30(N 79) +H. influenzae 2 3

+RSV 2 3 +Other(s) 8 10H. influenzae alone 2 3 +HHV-6 1 1 +RSV 1 1M. catarrhalis alone 1 1+Parainfluenza 3 1 1C. trachomatis 5 6M. pneumoniae 1 1None 31 39

Other respiratory 22 73 S. pneumoniae alone 4 13infection +M. catarrhalis 1 3(N 30) +Adenovirus 1 3

+RSV 1 3 +Other(s) 2 7S. aureus alone 3 10S. pyogenes 3 10H. influenzae type b 1 3 +C. trachomatis 1 3Adenovirus 3 10Other(s) 2 7None 8 27

Sepsis-like infection 25 68 S. pneumoniae alone 7 19(N 37) +Other(s) 2 5

H. influenzae 4 11M. catarrhalis 2 5Adenovirus 4 11Other(s) 6 16None 12 32

Other acute infection 10 42 S. pneumoniae alone 6 25(N 24) +Other(s) 2 8

S. aureus 1 4Influenza b 1 4None 14 58

Table 8. Etiologic agents identified in various clinical manifestations (I).

49

Figure 3. The etiologic spectrum in various clinical manifestations among the caseswith an identified etiology (N 105) (I).

8.2. Patients with randomized treatment (II, III)

8.2.1. Characteristics and bacterial etiology (II, III)

Initial characteristics of the 154 evaluable patients are presented in Table 9.Most of the patients had high fever, together with significantly increased CRP,WBC, and ESR values. On admission, in 80 % patients, CRP was ≥ 80 mg/L,while in 9 % of the cases CRP increased by 50 % in the following 6-8 hours.Microbiologic evidence for a bacterial etiology was obtained in altogether 55 %of the cases. (Table 10). Serologic evidence for S. pneumoniae infection wasfound in 59 cases and for Chlamydia or Mycoplasma infection in 13 cases.

On admission, the patients in the penicillin group tended more often to havecough and diarrhea than the children in the cefuroxime group. Vomiting andpneumonia was found more often in the penicillin group and sepsis-like diseasein the cefuroxime group. Otherwise, the treatment groups were almost identical(Table 9).

The symptoms and signs of the children in the 4-day and 7-day groups weresimilar on admission, although mixed bacterial-viral infections were morecommon in the children in the 4-day group (14 % vs. 2 %). In contrast, singlebacterial infections were more common in the 7-day group (47 % vs. 27 %).

50

Study II (Pen vs. Cfm) Study III (4- vs. 7-day)

Pen (N 75) Cfm (N 79) 4-day (N 73) 7-day (N 81) All patientsMean/ % Range Mean/ % Range Mean/ % Range Mean/ % Range Mean/%

Girls 43 46 49 40 44Boys 57 54 51 60 56Age (yrs) 3.9 0.7-11.8 3.9 0.3-14.0 3.8 0.7-14 3.9 0.3-12.7 1.4

Clinical signs / symptomsBody temperature (°C) 39.2 36.2-40.8 39.1 36.7-41 39.1 36.2-40.9 39.2 36.7-41 39.1Cough (%) 61 44 51 54Vomiting (%) 34* 18* 25 25Diarrhea (%) 9 5 7 8

Pretreatment 9 10 11 9antimicrobials (%)

Laboratory parametersWBC (x109/L) 21.7 3.6-42.5 22.3 8.3-45.2 21.6 3.6-45.2 22.4 9.0-44.4 22ESR (mm/H) 53 8-123 55 2-131 59 2-131 50 5-115 55CRP (mg/L) 137 23-400 127 12-334 131 33-270 132 12-334 132CRP ≥ 80 mg/L (%) 83 77 82 78 80CRP 50% increase 7 11 8 10 9in 8-12 hours (%)**

Disease manifestationPneumonia 55* 39* 46 47 47

Other respiratory 15 18 15 17 16infectionSepsis-like infection 17* 28* 25 21 23Other acute infection 13 15 14 15 14

* p < 0.05** Initial CRP value < 80 mg/L.

Table 9. Initial characteristics and disease manifestations in the 154 evaluablepatients in the SE-TU studies (II, III).

39.1

51

Table 10. Bacterial etiology of the 154 patients in the SE-TU studies (II, III).

Study II (Pen vs. Cfm) Study III (4- vs. 7-day)

Pen (N 75) Cfm (N 79) 4-day (N 73) 7-day (N 81) Total% % % % %

Bacteria in all 51 58 49 59 55single 39 37 27* 47* 38dual / triple 4* 14* 8 10 9mixed with virus(es) 8 8 14* 2* 8

Etiology disclosed in all 56 68 62 63 62

Specific agents N N N N TotalS. pneumoniae alone 20 19 15 24 39

with virus 3 6 7 2 9with Chlamydia / Mycoplasma 1 4 2 3 5with H. influenzae / 2 2 1 3 4M. catarrhaliswith other bacteria 0 2 0 2 2

H. influenzae 2 5** 1*** 6*** 7with virus 2 0 2 0 2with Chlamydia 0 11 0 1

M. catarrhalis alone 2 1 0 3 3with virus 1 0 1 0 1with Chlamydia 0 1 1 0 1

Chlamydia / Mycoplasma 2 4 3 3 6

Other bacteria 3 1 2 2 4

* p < 0.05** 2 type b*** 1 type b

8

11

52

8.2.2. Antimicrobial therapy, length of treatment, andrecovery (II, III)

Seventy-five patients were treated with penicillin, and 79 with cefuroxime (Table11). Clinical recovery and duration of fever or the need for antipyretics in the twogroups were similar (Table 11). Acute host response parameters, such as CRP,WBC, and ESR, normalized at equal rates during treatment (Figure 4). Nodifference was found in potentially drug-related adverse effects, including diarrhea,nausea, and eczema (Figure 5). Most children suffering from nausea in the penicillingroup had this symptom already on admission, before starting the medication(data not shown). In two patients initially treated with penicillin, the scheduledantimicrobial regimen was changed during treatment, and one patient in eachtreatment group was rehospitalized within 1-2 days, and a different antimicrobialwas instituted (Table 11).

The randomized length of treatment was 4 days in 73 cases, and 7 days in 81cases. Medication was completed parenterally as scheduled in 97 % of the patientsin the 4-day group, and in 62 % of those in the 7-day group. The remaining

Pen 4 Pen 7 Cfm 4 Cfm 7N 36 N 39 N 37 N 42

Realization Treatment completed parenterally 34 28 37 22Treatment completed orally 0 11 0 20

Treatment changed during 2 0 0 0treatment

Recovery Fever (> 37.5°C) / required antipyreticsday 2 17 (47%) 15 (38%) 12 (32%) 15 (36%)day 4 4 (11%) 6 (15%) 3 (8%) 6 (14%)

CRP, mean (mg/L)day 2 141 163 141 141day 4 50 57 53 46

Outcome Treatment completed but 0 1 1 0rehospitalization with anotherantimicrobial in 1-2 days

Patient rehospitalized within 1 month 1 0 0 1

Visited outpatient clinic / 3 4 3 4private physician in 1 month

Table 11. Realization of the randomized treatments, recovery, and outcome ofpatients in the SE-TU studies (II, III).

53

Figure 4. WBC, CRP, and ESR of patients in the Pen vs. Cfm study (II). Values arepresented as means with standard errors, p = NS.

Recovery during treatment was similar in the two groups, with no difference insymptoms or signs, or rate of normalization of laboratory parameters (data notshown). In two patients in the 4-day group, the randomized length of treatmentwas prolonged with another regimen, while one patient in each group wasrehospitalized within 1-2 days, and antimicrobial was reinstituted (Table 11).

The results were similar, when the data was looked between the four subgroupswith different regimens and different lengths of treatment used (Table 11).

patients in the latter group completed their medication orally at home (Table 11).

54

Figure 5. Symptoms and signs during recovery in the Pen vs. Cfm study, p = NS (II).

8.2.3. Clinical outcome (II, III)

Realization of the randomized treatments and outcome of the patients arepresented in Table 11. In two cases, the randomized antimicrobial was changedduring treatment and the course of therapy was prolonged. The first case was a2-year-old girl with periorbital cellulitis randomized to the 4-day penicillin group.Because of her slow response to procaine penicillin, the antimicrobial waschanged on the third day to cefuroxime for an additional four days. The etiologyremained unknown.

The other case was a 5-year-old girl with pneumonia, also randomized tothe 4-day penicillin group. During hospitalization, treatment was changed toerythromycin. Blood culture was negative, but C. trachomatis infection wasdiagnosed by serology. In both cases, penicillin treatment was consideredineffective.

Two children completed their treatment as scheduled, but were rehospitalizedwithin 1-2 days. A 4-year-old girl with bilateral pneumonia was randomized tothe 4-day cefuroxime group. The initial response was good but two days afterthe end of therapy, fever and respiratory symptoms reappeared. The child wasreadmitted and treated with parenteral penicillin. S. pneumoniae and HHV-6were identified by serology.

The other case was a 2-year old boy with bacteremic pneumococcal pneumonia(penicillin MIC < 0.1 µg/mL). After procaine penicillin for seven days, fever

55

and respiratory symptoms reappeared. Blood culture was negative and cefuroximewas administered for five days. Both viral antigen detection from NPA obtainedon first admission and viral serology from paired sera were negative. The symptomsmay have been caused by an undetected secondary infection.

Neither the antimicrobial used nor the length of therapy influenced thefrequency of visits to the outpatient clinic or some other physician within onemonth after hospitalization (Table 11). All the patients recovered.

8.3. Lung tap (IV, V)

8.3.1. Sample validity and overall yield

Lung tap was performed on 34 children, in 26 of whom an accurate sample withleukocytes was obtained. In most cases, the volume of the lung aspirate rangedfrom a few drops to a few milliliters. The lung aspirate disclosed the etiology in59 % of the cases. There were 16 single bacterial infections, two dual bacterialinfections, one mixed bacterial-viral infection, and one viral infection. Amongthe 26 patients with a representative sample, the etiology was established in69 %. Patients with a positive lung aspirate were significantly younger (mean 3.6vs. 7.0 years, p = 0.02), and had a higher CRP (216 mg/L vs. 134 mg/L, p= 0.01)and ESR (87 mm/H vs. 63 mm/H, p = 0.01) on admission than those with anegative sample. However, there were no differences either in the WBC (20.6 x109/L vs. 17.1 x 109/L) on admission, or in the duration of symptoms, or degreeof fever on admission (data not shown).

8.3.2. Etiology of consolidated pneumonia

S. pneumoniae was detected in the lung aspirate in 53 % of the cases, in 15cases as a single pathogen, once with enterovirus, once with C. pneumoniae,and once with Moraxella osloensis. M. pneumoniae pneumonia and viralpneumonia (RSV or parainfluenza) were each detected once (Table 12). Bloodculture was positive for S. pneumoniae only twice (Table 12).

RSV was detected from the nasopharynx in three children, with simultaneousisolation of S. pneumoniae by lung tap. In these cases, no viruses were foundfrom the lung tissue. In one case, conventional methods suggested S. pyogenestonsillitis without concomitant finding from lung aspirate. S. pneumoniae wastwice identified simultaneously from blood and lung.

56

Combining all the above-mentioned diagnoses, lung tap increased the proportionof patients with evidence of etiology from 15 % (5/34) to 62 % (21/34). Theproportion of cases in which a definitive etiology was established increased from6 % (2/34) to 59 % (20/34).

Table 13 shows the age, CRP, WBC, and body temperature of the patientswith a positive lung aspirate. All the parameters varied widely, regardless of theetiology identified. The radiographic findings caused by the different agentsidentified by lung tap are presented in Figure 6. Radiographs are presented fromthe cases in which different etiologies were identified, but, for pneumococcalpneumonia, only one typical chest radiograph was chosen (Figure 6A).

8.3.2.1. Moraxella osloensis (IV)

An atypical pathogen, M. osloensis, was isolated by lung tap from a 6-year-oldgirl with a 10-day history of cough and a 4-day history of abdominal pain. On

Lung Lung Blood Pnc** Pnc-RNA Hi** Hi-RNA C.pn** M.pn** Rhino Entero Viralaspirate aspirate culture PCR hybridi- PCR hybridi- PCR PCR PCR PCR culturestaining* culture zation zation

leuk+++ Pnc - + + - ND - - - - -leuk+++ Pnc - + + - ND - - - - -leuk+++ - - - ND - ND - - - - +***leuk+++ Pnc - + + - - - - - - -leuk+++ Pnc - ND + ND - ND ND ND ND -leuk++**** Pnc - + + - - - - - - -leuk++**** Pnc - + + - - - - - - -leuk++**** Pnc - + + - - - - - - NDleuk++ Pnc - - + - - - - - - -leuk++ - - + - - - - - - - -leuk++ Pnc - + + - - - - - + -leuk++ - - + - - - - - - - -leuk++ - - + - - - - - - - -leuk++ M. osloensis - + - - - - - - - -leuk+**** Pnc Pnc + + - - - - - - -leuk+ - - + ND - ND - ND ND ND -leuk+ - - + - - - + ND ND ND -leuk+ - Pnc + - - - - - - - -- - - - - - - - + - - -- - - + - - - - - - - -

* -, no leukocytes; +, 1-2 leukocytes; ++, 3-5 leukocytes; +++, >5 leukocytes seen with 100x magnification.** Pnc, S. pneumoniae; Hi, H. influenzae; C.pn, C. pneumoniae; M.pn, M. pneumoniae.*** Parainfluenza or RS virus, final typing impossible.**** Gram+cocci in staining

Table 12. Positive results obtained by different methods in the 20 cases in whichthe etiology was detected.

57

admission, she had fever (37.6 °C, a few hours later 39.7 °C), respiratory rate48 /min, and oxygen saturation 95 %. The CRP value was 260 mg/L, WBC21.8 x 109/L, and ESR 127 mm/H. A chest radiograph revealed bilateralconsolidation in the lower lobes (Figure 6B). Urine analysis, throat culture,viral antigen detection from NPA, and blood culture were negative, whileculturing of the needle yielded a pure culture of penicillin-sensitive M. osloensis.A likely concomitant pneumococcal infection was diagnosed afterwards by PCR,pneumococcal DNA-RNA hybridization still remaining negative.

8.3.3. Adverse eventsIn the check-up radiograph, pneumothorax was detected in six patients; fivewere considered iatrogenic, of whom one had concomitant pneumomediastinum.All these cases resolved spontaneously without any treatment or pleural drainage.The hospital stay and the length of antimicrobial therapy were not prolonged.Neither the length of the history, the pretap symptoms nor the signs or values of

Patient Age CRP WBC T Pathogen found Methodyears mg/L 109/L °C

4 2.9 279 23.2 39.5 Pnc culture5 0.8 264 28.6 40.5 Pnc culture7 8.7 234 25.8 39 RSV/parainfluenza culture8 12 220 17.7 39.7 Pnc culture9 1.1 102 17.2 38 Pnc PCR11 1.6 314 8.2 40 Pnc culture12 3 195 33.1 37.3 Pnc, C. pneumoniae PCR13 6 16 27.9 38.9 Pnc PCR14 1.5 131 24.3 39.2 Pnc culture15 1.8 345 22.9 37.3 Pnc culture17 1.3 186 20.9 38.3 Pnc culture19 14.8 222 6.0 39.4 M. pneumoniae PCR20 1 290 25.7 40 Pnc PCR21 2.5 330 14.3 38 Pnc, enterovirus culture, PCR22 1.8 144 17.7 36.6 Pnc PCR23 3.2 204 13.5 38 Pnc culture24 2.1 174 19.4 39.7 Pnc PCR27 4 196 15.5 38.2 Pnc PCR28 6.1 260 21.8 37.6 M. osolensis, Pnc culture, PCR29 1.3 232 29.4 38.2 Pnc culture

Mean 6.8 217 20.7 38.7

Range 0.8-14.8 16-345 6-33.1 36.6-40.5

Table 13. Age, C-reactive protein (CRP), white blood cell count (WBC) , andbody temperature on admission in patients in whom the etiology was identifiedby lung tap (IV, V).

58

Figure 6. Chest radiographs of patients with etiology disclosed by lung tap. An exampleof typical pneumococcal pneumonia (A) and all the other cases, in which differentetiologies were identified, are presented (B-F).

A 2.9-year-old boy; fever for 1 day. S. pneumoniae

B 6.1- year-old girl; cough for 10 days, abdominal pain for 4 days.M. osloensis+ S. pneumoniae

59

C 3.0- year-old boy; fever and vomiting for 4 days.S. pneumoniae + C. pneumoniae

D 2.5- year-old girl; fever, cough, vomiting for 2 days.S. pneumoniae + enterovirus

Figure 6. Continued.

60

E 14.8- year-old boy; fever and cough for 5 days. M. pneumoniae.

F 8.7- year-old girl; fever, vomiting, headache, and pain on right side ofthorax for 2 days. RSV or parainfluenza virus.

Figure 6. Continued.

61

acute-phase parameters on admission distinguished the patients with and withoutpneumothorax (data not shown). Nor did the frequency of obtaining a validsample correlate with that of post-tap pneumothorax (data not shown).

8.4. Diagnostics of pneumococcal pneumoniawith different methods (I, V)

Study I comprised 79 children with evidence of alveolar pneumonia; 36 caseswere caused by S. pneumoniae alone or concomitantly with other agents (Table8). Blood culture was positive in four cases. Serology alone provided 32diagnoses, while serology and blood culture were simultaneously positive inthree cases, and blood culture alone in one case (Table 14). In 66 % (23/35) ofthe serologically diagnosed infections, only one test was positive, while in 34 %(12/35) of the cases two tests were positive. All three assays were never positivesimultaneously. Fourty-six percent (16/35) of positive cases were identified byICs only while in 40 % (14/35) only free Ply antibodies were found.

The second series (V) included 34 children with consolidated pneumonia.Pneumococcus was detected from blood cultures in two cases, and from lungaspirates in 53 % (18/34) of patients, or in 65 % from those 26 of whom arepresentative aspirate sample was obtained. Of all the 18 pneumococcal diagnosesfrom lung aspirates, 10 were based on culture and eight were concomitantlypositive with Ply-PCR. PCR only detected eight additional cases (Table 12).

The 32 patients with only serology-proven pneumococcal pneumonia tendedto be older than those 22 in whom the agent was detected from blood culture orlung tissue (mean 4.7 vs. 3.3. years, p = NS). The former group had lowervalues for CRP (mean 144 mg/L vs. 201 mg/L, p = 0.02) (Figure 7) and ESR(66 mm/H vs. 84 mm/H, p = 0.04) on admission. There were no significantdifferences in WBC or the degree of fever (data not shown).

Assay* Ply-ab Ply-IC C-PS -IC Blood culture Total positive

Pneumolysin antibodies (Ply-ab) 12** 4 1 2 19Pneumolysin immune complex (Ply-IC) 4 8 7 1 20C-polysaccharide immune complex (C-PS -IC) 1 7 0 0 8Blood culture 2 1 0 1 4

* All assays performed on 79 patients.** Number of positive cases.

Table 14. Comparison of serologic methods and blood culture in pneumococcalpneumonia (I).

62

8.5. Pneumococcal serotypes and antimicrobialsusceptibility (V)

Twelve invasive S. pneumoniae strains were serotyped (10 lung and two bloodisolates). The following serotypes were found: type 3, 6A, 6B, 7F, 14, 18C,19A. Once the same serotype (18C) was detected from lung and blood.S. pneumoniae was found only from lung in five cases, while in six cases itwas simultaneously isolated from oropharyngeal samples and blood or lung.In one case different serotypes (19A and 23F) with different antimicrobialsensitivities were detected from lung and oropharynx. Three blood or lungisolates and one isolate from the oropharynx were intermediately resistant topenicillin (MIC < 1 µg/mL). Three lung strains were highly (MIC > 256 µg/mL)resistant to erythromycin.

Figure 7. CRP of patients with pneumococcal pneumonia diagnosed by blood cultureor lung tap vs. those detected by serology. Means are denoted with lines (p < 0.05).

63

9 DISCUSSION

9.1. Etiology of pneumonia and other suspectedinvasive infections warranting hospitalization (I)

For accurate antimicrobial treatment, causative pathogens should be identified.In this study, the etiology of pneumonia and other likely bacterial infections of170 children requiring parenteral antimicrobials and hospitalization were studiedprospectively.

9.1.1. Study designThe 170 children enrolled in the study were admitted to the on-call service ofeither the Aurora Hospital or the Children’s Hospital, University of Helsinkiduring two time periods, 1988 - 1989 and 1991 - 1993. Aurora Hospital servedas a referral hospital for the patients of practitioners of the Health Centers ofthe city of Helsinki, and as the only self-referral hospital for children in need ofhospital care living in the city of Helsinki. The Children’s Hospital was mainlyresponsible for children living in areas close to Helsinki.

The physician on duty examined the patients and made the decision for theneed of parenteral antimicrobial. Enrollment was based mainly on clinicalmanifestations suggestive of invasive bacterial infection. Detailed instructionsbased on CRP values were also formulated for the inclusion criteria in clinicallydisputable cases.

The aim of this prospective study was to collect all the patients fulfilling theinclusion criteria. It is possible, however, that some of the patients were notasked to participate. Furthermore, it is not known how many of the patientsrefused, when invited to participate in the study. Thus, the total number ofpatients who should have been enrolled is not known.

The heterogeneity of the patient material may be criticized, but the studywas planned to resemble the “every-day situation” of on-call service, wherevarious clinical manifestations have to be treated without knowledge of theetiology. The patients included in the study are a representative sample of thosechildren with a suspected invasive bacterial infection who are admitted to hospitalfor parenteral antimicrobials. Notably, the study does not give a reliable pictureof the etiology of childhood infections treated at home. Moreover, in this hospital-based study, infants and young children are over-represented, and thus the resultsare not directly applicable to school-aged children.

64

9.1.2. Microbiologic methods

Since few of the children had a clear focus of infection from which a representativemicrobiologic sample could have been obtained, the main diagnostic methodwas chosen to be serology. Paired serum samples were available from all thepatients. However, in some cases there was not enough serum for all the analyses,which may have reduced the number of cases with identified etiology.Interpretation of the bacteriologic results in this particular study was based onthe assumption that a significant antibody response or the presence of specificICs is indicative of an infection caused by the bacterium in question. However,the sensitivity and specificity of bacterial antibody assays are not well establishedfor the lower respiratory infections of childhood.

There are several factors that may have influenced the serologic results ofthis study. False-negative results because of diminished antibody responses maybe due to the young age of the child (Claesson et al. 1989, Nohynek et al. 1995,Korppi and Leinonen 1998), to too short a time from the onset of infection, toprevious antimicrobial therapy, to treatment begun soon after the onset ofinfection (Pichichero et al. 1987), or to too short an interval between the serumsamples. In this study, paired serum samples were obtained about two to threeweeks apart, which may have decreased the rate of positive findings especiallyin chlamydial MIF (Ekman et al. 1993). Furthermore, the assays used may havebeen suboptimal in respect of sensitivity. In the absence of a purified, species-specific protein antigen for EIA, a suboptimal whole-cell suspension of a mixtureof H. influenzae or M. catarrhalis isolates had to be used (Leinonen et al. 1981).

The pneumococcal etiology was investigated by measuring antibodies to Plyand ICs containing IgG antibodies to Ply, and C-PS. Since many children weretoo young to respond to most pneumococcal polysaccharides (Claesson et al.1989, Korppi and Leinonen 1998), antibodies to capsular polysaccharides werenot measured. In contrast to species-specific antibodies, measurements of theresponse to Ply and C-PS should theoretically detect all pneumococcal infections.However, it is known that all the assays for detecting pneumococci have limitedsensitivity, and consequently the absence of measurement of free antibodies toC-PS and of the response to type-specific capsular polysaccharides may havedecreased the number of pneumococcal infections detected (Korppi et al. 1993,Korppi and Leinonen 1998, Toikka et al. 1999b, Heiskanen-Kosma 2000).Moreover, some false- positive cases may have resulted from concomitant upperrespiratory infection (Virolainen et al. 1996, Rapola et al. 2001). Despite theseshortcomings of the specific tests and of serology in general, the assays used arethe most common standardized serologic tests currently available.

In this series, in accordance with previous studies (Nohynek et al. 1991,Juvén et al. 2000), all viral findings, whether reached by antigen or antibody

65

assays, were considered indicative of viral infection. However, wholly convincingevidence that a virus isolated from the upper respiratory tract is responsible forconcomitant lower respiratory infection is still mostly lacking (Hughes et al.1969, Avila et al. 1990). Since the primary goal of this study was to search forthe bacterial etiology of infections, the latest, more expensive virologic methodswere not utilized.

9.1.3. Etiologic findings in relation to previous studiesThe results of this series suggest considerable heterogeneity of the causativeagents in all the clinical manifestations studied. In all, evidence for an infectiousagent was obtained in 62 % of cases. In all entities, the agent most frequentlyfound was, as anticipated, S. pneumoniae, because this pathogen was searchedfor with three different serologic assays. In 21 of 62 cases, there was co-infectionwith another bacterium or virus. In this series, a purely viral etiology was rare,as was to be expected from the study design; all the patients included werehospitalized for parenteral antimicrobials. Among the patients with alveolarpneumonia, no viral infections were found but a mixed viral-bacterial etiologywas detected in 11 % of cases, or 19 % of those with evidence of a microbiologicetiology. However, in the sepsis-like infections (CRP ≥ 100 mg/L, fever ≥ 38.5°C,no respiratory symptoms), viruses alone were found in 19 % of the patients, andin 28 % of those in whom the etiology was identified. These cases were causedby adenovirus, which has already previously been shown to cause such highCRP values (Ruuskanen et al. 1985).

Comparison of the present results with those of previous studies is difficultbecause of lack of a similar study design. Most studies have addressed only onespecific manifestation. In most serology-based studies of pneumonia inhospitalized children, the proportions with viral and with mixed viral-bacterialetiology have been higher than in this series (Pailsey et al. 1984, Nohynek et al.1991, Ruuskanen et al. 1992, Juvén et al. 2000). This probably depends partlyon the lack of sensitive viral serology in this study. Furthermore, in this study,group pneumonia included only those cases with signs of alveolar pneumonia inthe radiograph. Other types of radiographic change, including probable interstitialpneumonia, were included in the group of other respiratory infections. A findingnot in accord with other serologic studies of the etiology of pneumonia (Heis-kanen-Kosma et al. 1998, 1999, Juvén et al. 2000), was that C. trachomatiswas found more commonly than C. pneumoniae. The reason for this is unknown.One explanation could be that, even though MIF is sensitive in species-specificdiagnosis when interpreted by an experienced reader, there still might be cross-reactivity with other types of Chlamydiae (Ozanne and Lefebvre 1992).

66

In this series, 66 % of cases with serologically proven pneumococcalpneumonia were diagnosed by a single method. As in previous studies (Korppiand Leinonen 1998, Toikka et al. 1999a, Heiskanen-Kosma 2000), detection ofspecific ICs was more sensitive than measurement of the antibody responses frompaired sera. Also in line with several other serology-based studies (Nohynek etal. 1991, Heiskanen-Kosma et al. 1998, Wubbel et al. 1999), here, in aconsiderable proportion of the children, the etiology remained unknown (38 %),despite the wide diagnostic panel used. The methodologic problems discussedabove probably explain this failure, but it is also possible that some cases werecaused by agents that were not searched for in this study, such as Bordetellapertussis, coronavirus, and rhinoviruses (He et al. 1998, Juvén et al. 2000).

9.2. Effectiveness of antimicrobial treatment (II)

The advent of cephalosporins in the 1970s and 1980s led to major changes inantimicrobial treatment in many countries, with increasing costs of therapy.Cephalosporins are more active in vitro against H. influenzae and M. catarrhalisthan the commonly used penicillins. However, there is no clear evidence thatthis simultaneously demonstrates clinical superiority (Klugman 1996), although,in a series of lower respiratory infections in adults, cefuroxime led to a betteroutcome than ampicillin (Pines et al. 1981).

In this study, we compared the clinical efficacies of parenteral penicillin andcefuroxime in common childhood infections warranting hospitalization. Thegroup of patients was considered to be a representative sample of the previouslyhealthy children hospitalized for parenteral antimicrobials. Since the regimenwas randomized, no selection should have occurred between the groups. Theprimary endpoint for successful treatment was uneventful recovery. Duration offever and need for antipyretics were not used as separate outcome measures,because antipyretics are frequently given to children as analgesics. Although, inabout half of the patients, a bacterial agent was not proven, the clinicalpresentation and laboratory values were highly suggestive of an invasive infection,and initiation of antimicrobials would probably have been the usual practice inmost hospitals.

It may be argued that the power of the study was too low to detect a differenceif there was any. However, power calculations indicated that a total sample sizeof 142 patients would disclose a 20 % difference in the effectiveness of the twotreatments. The study design, in which both the regimen used and the length oftreatment were randomized, may have caused problems when analyzing thetwo main groups and when assessing the roles of these two factors in the cases

67

considered as treatment failures. Because of the small sizes of the subgroups,statistical analysis was not done. The subgroups were, however, similar withregard to initial characteristics, recovery, and outcome.

It may also be argued that antimicrobial resistance – especially in cases ofS. pneumoniae – is now more common than at the time when the study wasconducted. In Finland, the results are, however, still applicable, since the levelof resistance has remained low (Forward 1999). Even if the resistance situationwere to change, pneumococcal penicillin resistance should not be a problem inthe treatment of pneumonia and sepsis (Friedland 1995, Pallares et al. 1995,Klugman 1996). Moreover, the rate of β-lactamase producing H. influenzaeand M. catarrhalis strains has remained quite stabile during recent years,suggesting that the efficacy of penicillin is probably quite similar to that duringthe study.

The relevance of β-lactam therapy in children in general may be questionedbecause of the possibility of infection with M. pneumoniae or C. pneumoniae(Ruuskanen and Mertsola 1999). Clinical failures have previously been reportedwhen M. pneumoniae pneumonia was treated with β-lactams (Ruuskanen et al.1992, Gendrel et al. 1997). On the other hand, in childhood pneumonia,eradication of C. pneumoniae has occurred in children < 5 years of age duringtreatment with amoxicillin-clavulanate (Harris et al. 1998). In this series, adiagnostic increase in either Chlamydia or Mycoplasma antibodies was foundin 13 children, 12 of whom were treated successfully with parenteral β-lactam.These findings raise two questions:1) What is the true pathogenic role of Chlamydia and Mycoplasma in these

clinical manifestations, and2) Is the seroresponse measured caused by current specific or nonspecific

infection, or by some previous disease, by asymptomatic carriage, or bysome other factor only.

Although there has been no previous comparative study on similar infectionsrequiring parenteral antimicrobials in children, indirect evidence for the clinicaleffectiveness and safety of the narrow spectrum antimicrobials has been obtainedfrom prospective and retrospective analyses of prescription habits and theirinfluence on the outcome. In the United Kingdom, no increase was observed inthe clinical treatment failures when several doctors in a pediatric hospital beganto treat common infections with parenteral penicillin, instead of the previouslyused broader-spectrum antimicrobials (Clements et al. 2000). Furthermore, arecent study from Uruguay showed that children with suspected bacterialpneumonia treated with penicillin and its derivatives had a good outcome (Pírezet al. 2001).

68

9.3. Shortening of therapy (III)

At present, the tendency is to discuss which antimicrobial to use rather than toquestion whether the optimum length of treatment could be shorter than it usedto be. Short regimens, when possible, would have obvious advantages in termsof compliance and cost (Pichichero and Cohen 1997). Other potential benefitsare fewer side effects and visits by the physician, and increased patient complianceand less trouble for the parents. Potential disadvantages, on the other hand, aretreatment failure, relapse, or recurrence, and an increased risk of complications(Pichichero and Cohen 1997).

This study investigated shortening of the length of parenteral β-lactam therapyin lower respiratory infections and in other common invasive infections in 154children warranting hospitalization. When either procaine penicillin or cefuroximewas used, no difference in outcome was found among patients treated for 4 or7 days. All the patients recovered. With longer parenteral treatment, an evidentproblem with compliance was seen in this study, since 38 % of children in the7-day group preferred to complete their medication orally at home. All thepatients who completed their randomized course either parenterally or orallywere included in the analysis, since previous findings had also suggested thatsequential antimicrobial therapy with a change from parenteral antimicrobial tooral administration provides efficacy equivalent to a full course of parenteraltreatment (Vogel 1995). In this series, there were no differences in the outcomeof the patients whose treatment was completed parenterally and those whosetreatment was completed orally at home (data not shown). Randomizing boththe regimen and the length of treatment did not either influence the outcome.

Since the study was conducted, general treatment polices regarding the lengthof parenteral therapy have changed. The mean length of parenteral administrationin the pneumonia series (V) of 34 children was only 3.5 days, although theantimicrobial was often continued with oral regimens. Recently, oral β-lactamtherapy for more than five days was shown to be associated with an increasedrisk of carriage of penicillin-resistant S. pneumoniae (Guillemot et al. 1998).Such a trend may be avoided by parenteral therapy (Klugman 1996). Theseearlier findings, together with the results of this study, suggest that 4-dayparenteral β-lactam therapy may be the most effective treatment in terms ofoutcome, compliance, and restriction of the spread of resistant strains.

69

9.4. Lung tap in the diagnosis of pediatricpneumonia (IV, V)

The difficulty of identifying the cause of pneumonia on clinical grounds, togetherwith the increasing number of bacteria resistant to the commonly usedantimicrobials has emphasized the importance of accurate etiologic diagnosis.In the absence of commonly used sensitive gold standard method, indirectmethods for detecting the etiology are mostly not optimally validated. In pneumonia,there is an evident need for improved diagnostics. This study tested the pertinenceof lung tap for this purpose. The aim was to obtain a sample directly from thefocus, uncontaminated by the oropharyngeal flora (Mimica et al. 1977).

9.4.1. Representativeness of the sample

This study included only patients with consolidated pneumonia. These childrenare usually treated in hospital, and their disease is usually relatively severe.Thus, the results do not give a reliable picture of childhood pneumonia in general.In patients treated as outpatients the disease tends to be milder, the radiographicchanges less clear, and a bacterial etiology may be less frequent. Lung tap isprobably not a sensitive method for such cases, since the yield improves withthe size of the consolidation (Zalacain et al. 1995).

9.4.2. Positivity rate of lung tapThe positivity rate of a lung tap depends on the representativeness of the sampleand on the microbiologic methods used. Since the location of the consolidationis determined from a chest radiograph and clinical signs only, there is always apossibility that the needle does not reach the consolidation, and that consequentlythe tap gives false-negative results. In this study, in accordance with a previousseries (Zalacain et al. 1995), the presence of leukocytes seen on staining wasconsidered indicative of a successful procedure.

Culture as such has limited sensitivity, which is known to be reduced somewhatby pre-tap antimicrobials (Vuori-Holopainen and Peltola 2001). Furthermore,the number of organisms present in the infectious focus may vary with the type

70

and with the phase of infection (Thomas and Parker 1920, Hammerschlag 1995).To avoid the effect of previous antimicrobials and too few pathogens presentfor culture, it may be valuable to use PCR and detection of antigens from thelung aspirates (Boersma et al. 1991, Ruiz-Gonzáles et al. 1997).

In this study, four bacterial and two viral PCR assays were used on the lungaspirates. These assays led to 10 diagnoses not detectable by other methods.Since PCR is extremely sensitive and does not require viable organisms, it maycause some false-positive results. Therefore, the relevance of a positive Ply-PCRin the absence of a simultaneous culture of pneumococcus may be questioned.A likely explanation is that the sample contained only a minute number of bacterialcells. Unlike previous studies of pneumonia applying Ply-PCR to blood samples(Rudolph et al. 1993, Salo et al. 1995, Dagan et al. 1998, Toikka 1999a), in thisstudy, each sample was taken directly from a consolidated infectious focus.Thus, in contrast to blood-based PCRs, false-positive results caused by naso-pharyngeal carriage (Dagan et al. 1998) should be rare. Furthermore, in thisstudy, a positive Ply-PCR was associated well with the presence of leukocytes inthe stained samples, suggesting that the assay had a high specificity. False-positiveresults caused by α-hemolytic streptococcus (Kearns et al. 2000) were evidentlynot a problem in this series since none of the lung aspirates grew that agent.

However, some unexpected negative results were obtained in cases with arepresentative sample. Some may have occurred as a result of previous anti-microbial treatment (Dagan et al. 1998). Another possible explanation is thatthe sample was taken too early (Scott and Hall 1999), and that therefore toofew pathogens were present (Thomas and Parker 1920). Finally, because thevolume of the samples was limited, no tests were made for the presence ofDNA polymerase inhibitors in PCR-negative samples (Higushi et al. 1989). Inoptimal settings, for elimination of false negative results, only patients withrepresentative sample should be included and tests should be made for thepresence of PCR inhibitors.

Like the present results, all previous studies employing lung tap have failedto prove the etiology in all cases (Scott and Hall 1999, Vuori-Holopainen andPeltola 2001). This may be due in part to the problems of methodology andsample validity discussed above. False-negative results are also obtained ifaspirate cultures are considered negative when an isolated organism does not fitwith our current knowledge of pneumonia etiology (Scott and Hall 1999). Ithas also been suggested that agents that still remain to be discovered, or arethought to be improbable pathogens may be responsible for some cases ofpneumonia without known etiology (Fang et al. 1990). In this study, M. osloensis

71

was identified once. A review of the literature removes any doubt of its pathogenicrole in pneumonia. One of the reasons why it has so seldom been identifiedbefore is probably that few pneumonia studies are based on culture, and noroutine PCR or serologic test for that bacterium has been available.

Another hypothesis to account for cases of pneumonia of unknown etiologyis that most of them are undetected pneumococcal infections (Ruiz-Gonzales etal. 1999). This view is based on the fact that study of lung aspirates has providedevidence of pneumococcal etiology in cases, which remained undiagnosed byconventional methods, and that sensitive PCR has further increased the numberof pneumococcal diagnoses. In this study, in accordance with that view, 16pneumococcal infections unrecognized by blood culture were detected by lungtap. Addition of other pneumococcal PCR assays, as well as antigen detectionmethods, might have further increased the numbers with pneumococcal etiology(Ruiz-Gonzales et al. 1997, Garcia et al. 1999).

9.4.3. Ethical considerations and applicability of lung tapInvasive methods, such as lung tap or transtracheal aspiration, are seldomconsidered justified, especially in developed countries, where the outcome ofpediatric pneumonia is favorable (Ruuskanen and Mertsola 1999). This view isbased on the belief that the risk of potentially harmful complications is considerable.

However, a recent review of the published literature on lung taps in childrenindicates that the method is safer than has been generally thought (Vuori-Holo-painen and Peltola 2001). With careful observation of the child after theprocedure, all potential adverse events requiring treatment can be dealt with.By restricting the use of the method to a few experienced physicians, the rate ofadverse events has been decreased, while the chances of obtaining a valid sampleare increased (Zalacain et al. 1995).

Lung tap has an undoubted role in etiologic studies of pneumonia. It providesconsiderable, highly specific, additional information on blood cultures, on speciesand serotypes causing disease, and on antimicrobial sensitivities. It is also avaluable methodologic tool for comparing other noninvasive diagnostic methods.Its role may become more important in the areas where antimicrobial resistanceis increasing. In view of these findings, and of the experience gained in thepresent study, we consider that lung tap is justified in well-equipped hospitals inresearch settings, and apart from the etiologic studies for patients who have notresponded to the initial therapy.

72

9.5. Invasive isolates of Streptococcuspneumoniae (V)

Identification of the pneumococcal serotypes causing most infections is criticalfor the design of multivalent conjugate vaccines. Since causative serotypes varyboth geographically and temporally (Butler et al. 1995, Sniadack et al. 1995),information is needed from several parts of the world. Pneumococci identifiedfrom the nasopharynx have been suggested as surrogate markers for those causingpneumonia or other forms of invasive pneumococcal disease (Lloyd-Evans etal. 1996). In this study, some degree of correlation was found between theoropharyngeal and the lung strains, but carriage without concomitant detectionfrom the lung was common, and, in one case, the lung and oropharynx yieldeddifferent types with different antimicrobial sensitivities. Furthermore, in fivecases, pneumococcus was isolated from the lung only. Current heptavalentconjugate vaccine (Rubin 2000) would have prevented only five of the 11 caseswith culture-positive pneumococcal pneumonia detected in this study. In future,type-specific pneumococcal PCR may provide additional information about theserotypes of the strains that are detectable only by sensitive PCR.

Although penicillin resistance has gained much attention worldwide, its rolein Finland has remained a minor one (Finres 1999). Our sample is small, and nodefinitive conclusions should be drawn. However, no fully penicillin-resistant strainswere found; meanwhile, three isolates were resistant (MIC > 256 µg/mL) toerythromycin. This is in accord with current resistance figures, which show that,in the Helsinki area, up to 12 % of S. pneumoniae strains are resistant to erythro-mycin (Dr M. Vaara, personal communication, 2001). This finding may reflect thecurrent trend for prescribing macrolides in childhood respiratory infections.

9.6. General discussion and futureconsiderations

The morbidity and mortality due to pneumonia have evidently been influencedby antimicrobials (Hui et al. 1997, Yang et al. 2001). The role of the new,widely-marketed, broad-spectrum agents, however, is controversial. In a retro-spective analysis of mortality from pneumonia in US children over the past half-century, the steepest decline in mortality occurred in the late 1940s, probably asa result of penicillin. A series of new antimicrobials has been introduced sincethen, but no association has been found between further reductions in mortalityand the introduction of these agents (Dowell et al. 2000).

73

In many nonindustrialized countries, high infant mortality is accepted byparents as a fact of life and is one reason for having large families. Markedreductions in childhood mortality due to pneumonia could probably be achievedby expanding immunization programs and by better diagnostic practices and byearly and appropriate management (Vaahtera et al. 2000). Inexpensive and easilyadministered medicaments would be especially useful in such settings (Kaplan2000). The present results with penicillin and with short courses of antimicrobialscannot directly be applied in the circumstances in developing countries, butthe experience may encourage the colleagues to conduct similar studies inthose countries.

In the industrialized world, the main reason for the indiscriminate use ofwide-spectrum antimicrobials is concern about bacteria potentially insensitiveto narrow-spectrum regimens. Besides comparative studies on the clinical efficacyof different antimicrobials, reliable information on the etiology is also requiredfor formulation of relevant guidelines for prescribing doctors (Clements et al.2000). On the basis of the present results, S. pneumoniae seem to be the mostcommon agent causing childhood pneumonia and other invasive bacterialinfections. Since the proportion of penicillin resistant strains has remained lowin our circumstances, penicillin can be considered as a treatment of choice inthese infections.

Priorities for future research on pediatric pneumonia include at least thefollowing:1) Determination of the sensitivity and specificity of commonly used, noninvasive

methods in relation to lung tap, a specific method for pneumonia diagnostics.2) Validation of various serologic methods in the light of the information

obtained by lung tap.3) Collecting data on pneumococcal serotypes obtained from lung tissue for

assessing the relevance of conjugate vaccine in the prevention of pneumonia.4) Re-evaluation of the relevance of acute-phase parameters and chest

radiographs for the etiologic diagnostics with respect to pathogens identifieddirectly from the infectious focus.

74

10 SUMMARY AND CONCLUSIONS

Although most cases of childhood community-acquired pneumonia and othercommon infections can be treated adequately with empirical medication, theetiologic diagnosis is important for seriously ill patients, and for those who havefailed to respond to standard therapy. Microbiologic diagnosis also enables theclinician to direct the treatment against specific pathogens and to avoidunnecessary medication. Furthermore, etiologic data are needed for treatmentguidelines and for development of vaccines.

In the present study, the etiology and treatment of childhood pneumonia,other respiratory infections, sepsis-like infections, and other acute infectionsrequiring parenteral antimicrobials were investigated. Meningitis, urinary tractinfections, and osteoarticular infections were excluded. Mainly serologic evidencefor a microbial etiology was found in 62 % of children. The distribution ofpathogens was heterogeneous, the dominant agent in all manifestations beingS. pneumoniae. These infections were cured by parenteral penicillin as well asby cefuroxime, and the 4-day parenteral β-lactam proved as effective as the 7-daytherapy. Hence, in most common childhood suspected invasive infections, shortcourses of narrow-spectrum parenteral β-lactam antimicrobials seem justified.

The accuracy of the microbiologic diagnostics in childhood consolidatedpneumonia requiring hospitalization was significantly improved by using a lungtap. Culture of the aspirate yielded the diagnosis in 32 % of 34 cases, but whenmodern microbiologic methods were applied, the etiology was identified in 59 %of all cases, and in 69 % of those in which a representative lung-tap specimenwas obtained. In consolidated pneumonia, the leading pathogen was S. pneumoniae.However, several other pathogens that were not predictable from the clinicalsigns or laboratory parameters of the patient were also found. The advantagesof lung tap are the high microbiologic yield and a relatively low risk of adverseevents. We consider lung tap applicable in hospital-based etiologic studies, andapart from research, in circumstances in which the need to identify the causativeagent outweighs the modest risks of the procedure.

75

11 ACKNOWLEDGMENTS

I wish to thank Emeritus Professor Jaakko Perheentupa, the former Head of theHospital for Children and Adolescents, University of Helsinki, and ProfessorMikael Knip, the current Head of the Hospital for Children and Adolescents,for providing excellent research facilities.

I am most grateful to my supervisor, Professor Heikki Peltola, for providingme with the study materials and facilities required and, more importantly, forteaching me all about pediatric infectious diseases, clinical research, science,and life in general. His vast experience and global view of infectious diseases,his innovative ideas and continuous energy, and, most of all, his never-failingencouragement and support have made this thesis possible.

I am grateful to all my collaborators, whose help has been invaluable duringthis work. I wish to express my warmest gratitude to Markku Kallio, MD, forhis enthusiastic guidance and advice in work as well as in personal life. Hisunlimited energy and his continuous disinterested support and help have beeninvaluable. I also warmly thank Eeva Salo, MD, and Docent Harri Saxén forcollecting the patients, for critical comments on the manuscripts, and for givingmuch advice about science, as well as for keeping my feet on the ground.Professor Martti Vaara, Professor Maija Leinonen, and Docent Klaus Hedmandeserve special thanks for their vast experience and valuable advice concerningour projects and manuscripts. I also want to thank all the other co-authors andthe SE-TU Study Group, for good and fruitful collaboration.

An essential part of this study has been collaboration with the nurses,laboratory personnel and other staff in the participating hospitals and laboratories.I owe my warmest thanks to all of them. Special thanks go to the personnel ofthe on-call service of Hospital for Children and Adolescents for whom our KE-TU study has sometimes caused plenty of extra work, and to Ms Raija Lahden-perä of Laboratory Diagnostics, Helsinki University Central Hospital, for allthe advice, e-mails, and phone calls during this project. Without her initiative,experience, and positive attitude, many things would have been much morecomplicated. I also want to express my gratitude to the patients and their parentsfor their co-operation, without which this study would not have beenaccomplished.

I would like to thank the official reviewers of this thesis, Professor JussiMertsola and Professor Matti Korppi, for their positive attitude, constructivecriticism, and good advice. Mrs Jean Margaret Perttunen revised the languageof this thesis, for which I am very grateful. I also highly appreciate Ms TerttuHalme´s work on the layout of the thesis.

I also acknowledge, with thanks, financial support of Helsinki UniversityBiomedical Graduate School, the EVO Funds of Helsinki University Central

76

Hospital, the Paulo Foundation, the Väinö and Laina Kivi Foundation, theResearch Foundation of Orion Corporation, the Finnish Medical Foundation,and Pfeizer Ltd. for this work.

Several colleagues and friends have given strong support during the project,which I highly appreciate. Tarja Heiskanen-Kosma, MD, has given valuablecriticism and unlimited support while I was writing the thesis. Annamari Mäkelä,MD and Päivi Sormunen, MD have started from the same “chick” phase withme and shared the process of how to do clinical research under the supervisionof Professor Heikki Peltola. Mrs. Riitta Meurman is warmly thanked for mentalsupport, for continuous interest in my family life, and for technical assistancewith the figures and tables. Colleagues Johanna Paronen and Kirsta Erkkilä arewarmly thanked for their friendship and for sharing the continuous difficultiesof being a researcher and a good mother.

My parents, Tuire and Arto, are thanked for their sympathy and support, andfor implanting in me the idea of the importance of meaningful work. My warmestthanks go to my sister Elisa. Her support, understanding, and friendship has somany times encouraged me to continue. I am also grateful to my parents-in-law, Pirkko and Martti, for all the support I have received.

My deepest gratitude goes to Juha, my dear husband. His endless love,support, encouragement, and understanding over the years have been invaluable.I could not have coped without him. And finally, I want to thank our deardaughter Ronja, who provides so much joy every day. You two, my belovedfamily, make my life worthwhile.

Helsinki, August 2001

77

12 REFERENCES

Abdel-Khalik AK, Askar AM, Ali M. The causative organisms of bronchopneumonia in infants inEgypt. Arch Dis Child 1938;13:333-342.

Adegbola RA, Falade AG, Sam BE, Aidoo M, Baldeh I, Hazlett D, Whittle H, Greenwood BM,Mulholland EK. The etiology of pneumonia in malnourished and well-nourished Gambianchildren. Pediatr Infect Dis J 1994;13:975-982.

American Thoracic Society. Guidelines for the initial management of adults with community-acquired pneumonia: Diagnosis, assesment of severity, and initial antimicrobial therapy.Am Rev Respir Dis 1993;148:1418-1426.

Amsden GW. Pneumococcal macrolide resistance – myth or reality? J Antimicrob Chemother1999;44:1-6.

Arason VA, Kristinsson KG, Sigurdsson JA, Stefánsdóttir G, Mölstad S, Gudmudssons S. Doantimicrobials increase the carriage rate of penicillin resistant pneumococci in children?Cross sectional prevalence study. Brit Med J 1996;313:387-391.

Ausina V, Coll P, Sambeat M, Puig I, Condom MJ, Luquin M, Ballester F, Prats G. Prospective studyon the etiology of community-acquired pneumonia in children and adults in Spain. Eur J ClinMicrobiol Infect Dis 1988;7:343-347.

Avila M, Salomón H, Carballal G, Ebekian B, Woyskovsky N, Cerqueiro MC, Weissenbacher M.Isolation and identification of viral agents in Argentinian children with acute lowerrespiratory infection. Rev Infect Dis 1990;12:S974-S981.

Bakaletz LO. Viral potentiation of bacterial superinfection of the respiratory tract. Trends inMicrobiol 1995;3:110-114.

Baquero F. Pneumococcal resistance to β-lactam antibiotics. Microb Drug Resist 1995;1:115-120.Baron EJ, Peterson LR, Finegold SM (ed.). Bailey and Scott’s diagnostic microbiology, 9th ed.,

p. 73-74. Mosby Year Book, St Louis, 1994.Bartlett JG, Mundy LK. Community-acquired pneumonia. N Engl J Med 1995;333:1618-1624.Berendt RF, Long GG, Walkers JS. Influenza alone and in sequence with pneumonia due to

Streptococcus pneumoniae in the squirrel monkey. J Infect Dis 1975;132:689-693.Berman S, McIntosh K. Selective primary health care: Strategies for control of disease in the

developing world. XXI. Acute respiratory infections. Rev Infect Dis 1985;7:674-691.Berman S. Epidemiology of acute respiratory infections in children of developing countries.

Rev Infect Dis 1991;13:S454-S462.Berner R. Back to the basics. Eur J Pediatr 2000;159:875-877.Boersma WG, Lowenberg A, Holloway Y, Kuttschrutter H, Snijder JA, Goeter GH. Pneumococcal

capsular antigen detection and pneumococcal serology in community-acquired pneumonia.Thorax 1991;46:902-906.

Briles DE, Yother J, McDaniel LS. Role of pneumococcal surface protein A in the virulence ofStreptococcus pneumoniae. Rev Infect Dis 1988;10:S372-S374.

Bromberg K, Hammerschlag MR. Rapid diagnosis of pneumonia in children. Semin Respir Infect1987;3:159-165.

Brook I, Yocum P. Bacteriology and beta-lactamase activity in ear aspirates of acute otitis media thatfailed amoxicillin therapy. Pediatr Infect Dis J 1995;14:805-808.

Burman LA, Leinonen M, Trollfors B. Use of serology to diagnose pneumonia caused bynonencapsulated Haemophilus influenzae and Moraxella catarrhalis. J Infect Dis1994;170:220-222.

Butler JC, Breiman RF, Lipman HB, Hofmann J, Facklam RR. Serotype distribution of Streptococcuspneumoniae infections among preschool children in the United States, 1978-1994:implications for development of a conjugate vaccine. J Infect Dis 1995;171:885-889.

Camargos PAM, Guimaraes MDC, Ferreira CS. Benzathine penicillin for unilateral lobar orsegmental infiltrates presumptively caused by Streptococcus pneumoniae in children 2-12years old. J Trop Pediatr 1997;43:353-360.

78

Campbell H, Byass P, Forgie IM, O’Neill KP, Lloyd-Evans N, Greenwood BM. Trial ofco-trimoxazole versus procaine penicillin versus procaine penicillin with ampicillin intreatment of community-acquired pneumonia in young Gambian children. Lancet1988;2:1182-1183.

Carbon C, Bax RP. Regulating the use of antibiotics in the community. Brit Med J 1998;317:663-665.Carson JL, Collier AM, Hu SS. Acquired ciliary defects in nasal epithelium in children with acute

viral upper respiratory infections. N Engl J Med 1985;312:463-468.Claesson BA, Trollfors B, Brolin I, Granström M, Henrichsen J, Jodal U, Juto P, Kallings I,

Kanclerski K, Lagergård T, Steinwall L, Strannegård Ö. Etiology of community-acquiredpneumonia in children based on antibody responses to bacterial and viral antigens. PediatrInfect Dis J 1989;8:856-861.

Clements H, Stephenson TJ. Blood culture is poor method of confirming pneumococcus as cause ofchildhood pneumonia. Brit Med J 1996;313:757.

Clements H, Stephenson T, Gabriel V, Harrison T, Millar M, Smyth A, Tong W, Linton CJ.Rationalised prescribing for community-acquired pneumonia: a closed loop audit. Arch DisChild 2000;83:320-324.

Cohen R, Levy C, Boucherat M, Lanques J, Aytret E, Gehanno P, de la Rocque F. Five vs ten daysof antibiotic therapy for acute otitis media in young children. Pediatr Infect Dis J2000;19:458-463.

Cunanan GD, Yambao JJ, Caces RF, Wellmonte MA, Rosales LM. Diagnostic lung aspiration insevere and complicated childhood pneumonia. Philip J Pediatr 1979;28:169-178.

Dagan R, Shirker O, Hazan I, Leibovitz E, Greenberg D, Schlaeffer F, Levy R. Prospective study todetermine clinical relevance of detection of pneumococcal DNA in sera of children by PCR. J Clin Microbiol 1998;36:669-673.

Davidson M, Tempest B, Palmer D. Bacteriologic diagnosis of acute pneumonia: comparison ofsputum, transtracheal aspirates, and lung aspirates. JAMA 1976;235:158-163.

Doern GV, Jones RN, Pfaller MA, Kugler K. Haemophilus influenzae and Moraxella catarrhalisfrom patients with community-acquired respiratory tract infections: antimicrobialsusceptibility patterns from SENTRY antimicrobial surveillance program (United States andCanada, 1997). Antimicrob Agent Chemother 1999;43:385-389.

Dowell SF, Kupronis BA, Zell ER, Shay DK. Mortality from pneumonia in children in the UnitedStates, 1939 through 1996. N Engl J Med 2000;342:1399-1407.

Du Clos TW. Function of C-reactive protein. Ann Med 2000;32:274-278.Ejlertsen T, Skov R. The β-lactamases of Moraxella (Branhamella) catarrhalis isolated from Danish

children. APMIS 1996;104:557-562.Ekman MR, Leinonen M, Syrjälä H, Linnanmäki E, Kujala P, Saikku P. Evaluation of serological

methods in the etiological diagnosis of Chlamydia pneumoniae pneumonia during anepidemic in Finland. Eur J Clin Microb Infect Dis 1993;12:756-760.

Eskola J, Takala AK, Kela E, Pekkanen E, Kalliokoski R, Leinonen M. Epidemiology of invasivepneumococcal infections in children in Finland. JAMA 1992;268:3323-3327.

Eskola J, Kilpi T, Palmu A, Jokinen J, Eerola M, Haapakoski J, Herva E, Takala A, Käyhty H,Karma P, Kohberger R, Lockhart S, Siber G, Mäkelä PH. Efficacy of a pneumococcalconjugate vaccine against acute otitis media. N Engl J Med 2001;344:403-409.

Fagbule DO. Bacterial pathogens in malnourished children with pneumonia. Trop Geogr Med1993;45:294-296.

Fainstein V, Musher DM, Cate TR. Bacterial adherence to pharyngeal cells during viral infection.J Infect Dis 1980;141:172-176.

Falade AG, Mulholland EK, Adegbola RA, Greenwood BM. Bacterial isolates from blood andlung aspirate cultures in Gambian children with lobar pneumonia. Ann Trop Paediatr1997;17:315-317.

Fang GD, Fine M, Orloff J, Arisumi O, Yu VL, Kapoor W, Grayston JT, Wang SP, Kohler R,Muder RR. New and emerging etiologies for community-acquired pneumonia withimplications for therapy. A prospective multicenter study of 359 cases. Medicine1990;69:307-316.

Farrington M, Rubenstein D. Antigen detection in pneumococcal pneumonia. J Infect 1991;23:109-116.

79

Fattom A, Schneerson R, Szu SC, Vann WF, Shiloach J, Karakawa WW, Robbins JB. Synthesis andimmunologic properties in mice of vaccines composed of Staphylococcus aureus type 5 andtype 8 capsular polysaccharides conjugated to Pseudomonas aeruginosa exotoxin A.Infect Immun 1990;58:2367-2374.

Fattom A, Schneerson R, Watson DC, Karakawa WW, Fitzgerald D, Pastan I, Li X, Shiloach J,Bryla DA, Robbins JB. Laboratory and clinical evaluation of conjugate vaccines composed ofStaphylococcus aureus type 5 and type 8 capsular polysaccharides bound to Pseudomonasaeruginosa recombinant exoprotein A. Infect Immun 1993;61:1023-1032.

Felmingham D, Robbins MJ, Tesfaslasie Y, Harding I, Shrimpton S, Gruneberg RN. Antimicrobialsusceptibility of community-acquired lower respiratory tract bacterial pathogens isolated inthe UK during the 1995-1996 cold season. J Antimicrob Chemother 1998;41:411-415.

Finland M. Diagnostic lung puncture. Pediatrics 1969;44:471-473.Finres 1999. Antibiotic resistance in Finland [in Finnish] p. 10. Ministry for Welfare and Health,

Helsinki, Finland, 1999.Forgie IM, O’Neill KP, Lloyd-Evans N, Leinonen M, Campbell H, Whittle HC, Greenwood BM.

Etiology of acute lower respiratory tract infections in Gambian children: I. Acute lowerrespiratory tract infection in infants presenting at the hospital. Pediatr Infect Dis J1991a;10:33-41.

Forgie IM, O’Neill KP, Lloyd-Evans N, Leinonen M, Campbell H, Whittle HC, Greenwood BM.Etiology of acute lower respiratory tract infections in Gambian children: II. Acute lowerrespiratory tract infection in children ages one to nine years presenting at the hospital.Pediatr Infect Dis J 1991b;10:42-47.

Forward KR. The epidemiology of penicillin resistance in Streptococcus pneumoniae. Semin RespirInfect 1999;14:243-254.

Foy HM, Cooney MK, Maletzky AJ, Grayston JT. Incidence and etiology of pneumonia, croup andbronchiolitis in preschool children belonging to a prepaid medical care group over a four-yearperiod. Am J Epidemiol 1973;97:80-92.

Foy HM. Infections caused by Mycoplasma pneumoniae and possible carrier state in differentpopulations of patients. Clin Infect Dis 1993;17:S37-S46.

Friedland IR, McCracken GH. Management of infections caused by antibiotic-resistant Streptococcuspneumoniae. N Engl J Med 1994;331:377-382.

Friedland IR. Comparison of the response to antimicrobial therapy of penicillin-resistant andpenicillin-susceptible pneumococcal disease. Pediatr Infect Dis J 1995;14:885-890.

Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl JMed 1999;340:448-454.

García A, Rosón B, Pérez JL, Verdaguer R, Dorca J, Carratala J, Casanova A, Manresa F, Gudiol F.Usefullness of PCR and antigen latex agglutination test with samples obtained bytransthoracic needle aspiration for diagnosis of pneumococcal pneumonia. J Clin Microbiol1999;37:709-714.

Gay K, Baughman W, Miller Y, Jackson D, Whitney CG, Schuchat A, Farley MM, Tenover F,Stephens DS. The emergence of Streptococcus pneumoniae resistant to macrolideantimicrobial agents: a 6-year population-based assessment. J Infect Dis 2000;182:1417-1424.

Gendrel D, Raymond J, Moulin F, Iniguez JL, Ravilly S, Habib F, Lebon P, Kalifa G. Etiology andresponse to antibiotic therapy of community-acquired pneumonia in French children. Eur JClin Microbiol Infect Dis 1997;16:388-391.

Gendrel D, Bohuon C. Procalcitonin as a marker of bacterial infection. Pediatr Infect Dis J 2000;19:679-687.

Goldmann DA, Weinstein RA, Wenzel RP, Tablan OC, Duma RJ, Gaynes RP, Schlosser J,Martone WJ, for the Workshop. Strategies to prevent and control the emergence and spreadof antimicrobial-resistant micro-organisms in hospitals. A challenge to hospital leadership.JAMA 1996;275:234-240.

Gray BM, Converse GM III, Dillon HC Jr. Epidemiologic studies on Streptococcus pneumoniae ininfants: acquisition, carriage, and infection during the first 24 months of life. J Infect Dis1980;142:923-933.

80

Grayston JT, Campbell LA, Kuo CC, Mordhorst CH, Saikku P, Thom DH, Wang SP.A new respiratory tract pathogen: Chlamydia pneumoniae strain TWAR. J Infect Dis1990;161:618-625.

Guillemot D, Carbon C, Balkau Bgreslin P, Lecoeur H, Vauzelle-Kervroedran F. Low dosage andlong treatment duration of β-lactam: risk factors for carriage of penicillin-resistantStreptococcus pneumoniae. JAMA 1998;279:365-370.

Gwaltney JM, Sande MA, Austrian R, Owen Hendley J. Spread of Streptococcus pneumoniae infamilies. II. Relation of transfer of S. pneumoniae to incidence of colds and serum antibody.J Infect Dis 1975;132:62-68.

Hammerschlag MR. Chlamydia trachomatis in children. Pediatr Ann 1994;23:349-353.Hammerschlag MR. Atypical pneumonias in children. Adv Pediatr Infect Dis 1995;10:1-39.Harris JAS, Kolokathis A, Campbell M, Cassell GH, Hammerschlag MR. Safety and efficacy of

azithromycin in the treatment of community-acquired pneumonia in children. Pediatr InfectDis J 1998;17:865-871.

He Q, Viljanen MK, Arvilommi H, Aittanen B, Mertsola J. Whooping cough caused byBordetella pertussis and Bordetella parapertussis in an immunized population. JAMA1998;280:635-637.

Hedman K, Lappalainen M, Söderlund M, Hedman L. Avidity of IgG in serodiagnosis of infectiousdiseases. Rev Med Microbiol 1993;4:123-129.

Heiskanen-Kosma T, Korppi M, Jokinen C, Heinonen K. Risk factors for community-acquiredpneumonia in children: a population-based case-control study. Scand J Infect Dis1997;29:281-285.

Heiskanen-Kosma T, Korppi M, Jokinen C, Kurki S, Heiskanen L, Juvonen H, Kallinen S, Stén M,Tarkiainen A, Rönnberg PR, Kleemola M, Mäkelä PH, Leinonen M. Etiology of childhoodpneumonia: serologic results of a prospective, population-based study. Pediatr Infect Dis J1998;17:986-991.

Heiskanen-Kosma T, Korppi M, Laurila A, Jokinen C, Kleemola M, Saikku P. Chlamydiapneumoniae is an important cause of community-acquired pneumonia in school-agedchildren: Serologic results of a prospective, population-based study. Scand J Infect Dis1999;31:255-259.

Heiskanen-Kosma T. Community-acquired pneumonia in childhood. A population-based study onbacterial etiology and risk factors. Academic Dissertation. University of Kuopio, 2000.

Henderson FW, Collier AM, Sanyal MA, Watkins JM, Fairclough DL, Clyde WAJ, Denny FW.A longitudinal study of respiratory viruses and bacteria in the etiology of acute otitis mediawith effusion. N Engl J Med 1982;306:1377-1383.

Higuchi R, Erlich HA (ed.). PCR technology: Principles and applications for DNA amplification,p.31-38. Stockton Press, New York, 1989.

Hjuler IM, Hansen MB, Olsen B, Renneberg J. Bacterial colonization of the larynx and trachea inhealthy children. Acta Paediatr 1995;84:566-568.

Hobson RP, Williams A, Rawal K, Pennington TH, Forbes KJ. Incidence and spread of Haemophilusinfluenzae on an Antarctic base determined using the polymerase chain reaction. EpidemiolInfect 1995;114:93-103.

Holloway Y, Snijder J, Boersma W. Demonstration of circulating immunoglobulin G immunecomplexes by means of an enzyme-linked immunosorbent assay. J Clin Microbiol1993;31:3247-3254.

Hughes JR, Sinha DP, Cooper MR, Shah KV, Bose SK. Lung tap in childhood. Bacteria, viruses,and mycoplasma in acute lower respiratory tract infections. Pediatrics 1969;44:477-485.

Hui L, Li XS, Zeng XJ, Fai YH, Foy HM. Patterns and determinants of use of antibioitics for acuterespiratory tract infection in children in China. Pediatr Infect Dis J 1997;16:560-564.

Imakita M, Shiraki K, Yutani C, Ishibashi-Ueda H. Pneumonia caused by rhinovirus. Clin InfectDis 2000;30:611-612.

Isaacman DJ, Karasic R, Reynolds EA, Kost SI. Effect of number of blood cultures and volume ofblood on detection of bacteremia in children. J Pediatr 1996;128:190-195.

Jadavji T, Law B, Lebel MH, Kennedy WA, Gold R, Wang EE. A practical guide for the diagnosisand treatment of pediatric pneumonia. Can Med Assoc J 1997;156:703-711.

81

Jalava J, Kotilainen P, Nikkari S, Skurnik M, Vanttinen E, Lehtonen OP, Eerola E, Toivanen P. Useof the polymerase chain reaction and DNA sequencing for detection of Bartonella quintanain the aortic valve of a patient with culture-negative infective endocarditis. Clin Infect Dis1995;21:891-896.

Jalava J, Eerola E. Phylogenetic analysis of Fusobacterium alocis and Fusobacterium sulci based on16S rRNA gene sequences: proposal of Filifactor alocis (Cato, Moore and Moore) comb. nov.and Eubacterium sulci (Cato, Moore and Moore) comb. nov. Int J Syst Bact 1999;49:1375-1379.

Jalonen E, Paton JC, Koskela M, Kerttula Y, Leinonen M. Measurement of antibody responses topneumolysin: a promising method for the presumptive etiological diagnosis of pneumococcalpneumonia. J Infect 1989;19:127-134.

Jarva H, Jokiranta TS, Hellwage J, Zipfel PF, Meri S. Regulation of complement activation byC-reactive protein: targeting the complement inhibitory activity of factor H by an interactionwith short concensus repeat domains 7 and 8-11. J Immunol 1999;163:3957-3962.

Jero J, Alakärppä H, Virolainen A, Saikku P, Karma P. Polymerase chain reaction assay for detectingChlamydia pneumoniae in middle ear fluid of children with otitis media with effusion.Pediatr Infect Dis J 1999;18:939-940.

Johnston RB. Pathogenesis of pneumococcal pneumonia. Rev Infect Dis 1991;13:S509-S517.Jokinen C, Heiskanen L, Juvonen H, Kallinen S, Karkola K, Korppi M, Kurki S, Rönnberg PR,

Seppä A, Soimakallio S, Stén M, Tanska S, Tarkiainen A, Tukiainen H, Pyörälä K,Mäkelä PH. Incidence of community-acquired pneumonia in the population of fourmuncipalities in eastern Finland. Am J Epidemiol 1993;137:977-988.

Julander IG, Granström M, Hedström SA, Molby R. The role of antibodies against alpha-toxin andteichoic acid in the diagnosis of staphylococcal infections. Infection 1983;11:77-83.

Juvén T, Mertsola J, Waris M, Leinonen M, Meurman O, Roivainen M, Eskola J, Saikku P,Ruuskanen O. Etiology of community-acquired pneumonia in 254 hospitalized children.Pediatr Infect Dis J 2000;19:293-298.

Kallio MJ, Unkila-Kallio L, Aalto K, Peltola H. Serum C-reactive protein, erythrocyte sedimentationrate and white blood cell count in septic arthritis of children. Pediatr Infect Dis J1997;16:411-413.

Kaplan EL. Whither penicillin? A response to the question. Pediatr Infect Dis J 2000;19:177-178.Kearns AM, Wheeler J, Freeman R, Seiders PR, Perry J, Whatmore AM, Dowson CG. Pneumolysin

detection identifies atypical isolates of Streptococcus pneumoniae. J Clin Microbiol2000;38:1309-1310.

Keeley DJ, Nkrumah FK, Kapuyanyika C. Randomized trial of sulfamethoxazole + trimethoprimversus procaine penicillin for the outpatient treatment of childhood pneumonia in Zimbabwe.Bull WHO 1990;68:185-192.

Kelley R, Keyserling H. Antigen excretion in the urine after pneumococcal vaccination. PediatrInfect Dis J 1997;16:720.

Kim PE, Musher DM, Glezen WP, Rodriguez-Barradas MC, Nahm WK, Wright CE. Association ofinvasive pneumococcal disease with season, atmospheric conditions, air pollution, and theisolation of respiratory viruses. Clin Infect Dis 1996;22:100-106.

King JC Jr, Lagos R, Bernstein DI, Piedra PA, Kotloff K, Bryant M, Choi I, Belshe RB.Safety and immunogenicity of low and high doses of trivalent live cold-adapted influenzavaccine administered intranasally as drops or spray to healthy children. J Infect Dis1998;177:1394-1397.

Kleemola SRM, Karjalainen JE, Räty RKH. Rapid diagnosis of Mycoplasma pneumoniae infection:clinical evaluation of commercial probe test. J Infect Dis 1990;162:70-75.

Kleemola M, Heiskanen-Kosma T, Nohynek H, Jokinen S, Korppi M, Eskola J. Diagnostic efficacyof a Mycoplasma pneumoniae hybridization test in nasopharyngeal aspirates in children.Pediatr Infect Dis J 1993;12:344-345.

Klein JO. The epidemiology of pneumococcal disease in infants and children. Rev Infect Dis1981;3:246-253.

Klugman KP. Pneumococcal resistance to antibiotics. Clin Microb Rev 1990;3:171-196.

82

Klugman KP. The clinical relevance of in-vitro resistance to penicillin, ampicillin, amoxicillin andalternative agents, for the treatment of community-acquired pneumonia caused byStreptococcus pneumoniae, Haemophilus influenzae and Moraxella catarrhalis.J Antimicrob Chemother 1996;38:133-140.

Kontiokari T, Renko M, Kaijalainen T, Kuisma L, Leinonen M. Comparison of nasal swab culture,quantitative culture of nasal mucosal tissue and PCR in detecting Streptococcus pneumoniaecarriage in rats. APMIS 2000;108:734-738.

Korppi M, Halonen P, Kleemola M, Launiala K. Viral findings in children under the age oftwo years with expiratory difficulties. Acta Pediatr Scand 1986;75:457-464.

Korppi M, Leinonen M, Mäkelä PH, Launiala K. Bacterial coinfection in children hospitalized withrespiratory syncytial virus infections. Pediatr Infect Dis J 1989;8:687-692.

Korppi M, Leinonen M, Mäkelä PH, Launiala K. Bacterial involvement in laryngeal infections inchildren. Acta Pediatr Scand 1990;79:564-565.

Korppi M, Kaitila ML, Jääskeläinen J, Leinonen M. Role of Moraxella (Branhamella) catarrhalis asa respiratory pathogen in children. Acta Paediatr 1992a;81:993-996.

Korppi M, Katila ML, Kalliokoski R, Leinonen M. Pneumococcal finding in a sample from upperairways does not indicate pneumococcal infection of lower airways. Scand J Infect Dis1992b;24:445-451.

Korppi M, Heiskanen-Kosma T, Jalonen E, Saikku P, Leinonen M, Halonen P, Mäkelä PH.Aetiology of community-acquired pneumonia in children treated in hospital. Eur J Pediatr1993a;152:24-30.

Korppi M, Heiskanen-Kosma T, Leinonen M, Halonen P. Antigen and antibody assays in theaetiologic diagnosis of respiratory infection in children. Acta Pediatr 1993b;82:137-141.

Korppi M, Kroger L, Laitinen M. White blood cell and differential counts in acute respiratory viraland bacterial infections in children. Scand J Infect Dis 1993c;25:435-440.

Korppi M, Leinonen M. Pneumococcal pneumonia in children: new data from circulating immunecomplexes. Eur J Pediatr 1997;156:341-342.

Korppi M, Leinonen M. Pneumococcal immune complexes in the diagnosis of lower respiratoryinfections in children. Pediatr Infect Dis J 1998;17:922-955.

Koskela M. Serum antibodies to pneumococcal C-polysaccharide in children: a response to acutepneumococcal otitis media or to vaccination. Pediatr Infect Dis J 1987;6:519-526.

Kozyrskyj AL, Hildes-Ripstein GE, Longstaffe SE, Wincott JL, Sitar DS, Klassen TP, Moffatt ME.Treatment of acute otitis media with shortened course of antibiotics.JAMA 1998;279:1738-1742.

van Kuppeveld FJM, van der Logt JTM, Angulo AF, van Zoest MJ, Quint WG, Niester HG,Galama JM, Mechers WJ. Genus and species-specific identification of Mycoplasmas by16S rRNA amplification. Appl Environ Microbiol 1992;58:2606-2615.

Lauer BA, Reller LB, Mirret S. Comparison of acridine orange and gram stains for detection ofmicroorganisms in cerebrospinal fluid and other clinical specimens. J Clin Micorbiol1981;14:201-205.

Lehtomäki K, Leinonen M, Takala A, Hovi T, Herva E, Koskela M. Etiologic diagnosis ofpneumonia in military conscripts by combined use of bacterial culture and serologicalmethods. Eur J Clin Microbiol Infect Dis 1988;7:348-354.

Leinonen M, Luotonen J, Herva E, Valtonen K, Mäkelä PH. Preliminary serologic evidence for apathologic role of Branhamella catarrhalis. J Infect Dis 1981;144:570-574.

Leinonen M, Syrjälä H, Jalonen E, Kujala P, Herva E. Demonstration of pneumolysin antibodies incirculating immune complexes - a new diagnostic method for pneumococcal pneumonia.Serodiagn Immunother Infect Dis 1990;4:451-458.

Leyden I. VII Verhandlungen des Vereins für Innere Medizin: über infectiose Pneumonie. DeutscMed Wschr 1882;4:52-54.

Lieberman D, Schlaeffer F, Boldur I, Lieberman O, Horowitz S, Friedman PG, Leinonen M,Horowitz O, Manor E, Porath A. Multiple pathogens in adult patients admitted withcommunity-acquired pneumonia: a one year prospective study of 346 consecutive patients.Thorax 1995;51:179-184.

83

Lin TY, Chrane SF, Nelson JD, McCracken GH. Seven days of ceftriaxone therapy is as effective asten days treatment for bacterial meningitis. JAMA 1985;253:3559-3563.

Linnavuori K, Peltola H, Hovi T. Serology versus clinical signs or symptoms and main laboratoryfindings in the diagnosis of exanthema subitum (Roseola infantum).Pediatrics 1992;89:103-106.

Lloyd-Evans N, O’Dempsey TJD, Baldeh I, Secka O, Demba E, Todd JE, Mcardle TF, Banya WS,Greenwood BM. Nasopharyngeal carriage of pneumococci in Gambian children and in theirfamilies. Pediatr Infect Dis J 1996;15:866-871.

Loo VG, Richardson S, Quinn P. Isolation of Mycoplasma pneumoniae from pleural fluid. DiagnMicrobiol Infect Dis 1991;14:443-445.

Lorente MLL, Falguera M, Nogués A, Ruiz Gonzáles A, Merino MT, Rubio Caballero M. Diagnosisof pneumococcal pneumonia by polymerase chain reaction (PCR) in whole blood:a prospective clinical study. Thorax 2000;55:133-137.

Lönnrot M, Sjöroos M, Salminen K, Maaronen M, Hyypiä T, Hyöty H. Diagnosis of enterovirus andrhinovirus infections by RT-PCT and time-resolved fluorometry with lanthadine chelatelabeled probes. J Med Virol 1999;59:378-384.

MacFarlane JT, Anjorin FI, Cleland PG, Hassan-King M, Tor-Agbidye S, Wali S, Weir WR,Whittle HC, Yahaya HN, Greenwood BM. Single injection treatment of meningococcalmeningitis. 1. Long-acting penicillin. Transact Royal Soc Trop Med Hyg 1979;73:693-697.

MacFarlane JT, Finch RG, Ward MJ, Rose DH. Erythromycin compared with a combination ofampicillin and amoxycillin as initial therapy for adults with pneumonia includingLegionnaires’ disease. J Infect 1983;7:111-117.

MacFarlane JT, Prewitt J, Gard P, Guion A. Comparison of amoxycillin and clarithromycin as initialtreatment of community-acquired lower respiratory tract infections. Brit J Gen Pract1996;46:357-360.

Marrie TJ. Community-acquired pneumonia. Clin Infect Dis 1994;18:501-515.McCarthy PL, Jekel JF, Dolan TF. Temperature greater than or equal to 40°C in children less than

24 months of age: A prospective study. Pediatrics 1977;59:663-668.McCarthy PL, Sharpe MR, Spiesel SZ, Dolan TF, Forsyth BW, DeWitt TG, Fink HF, Baron MA,

Cicchetti DV. Observation scales to identify serious illness in febrile children. Pediatrics1982;70:802-809.

McCracken GH Jr. Etiology and treatment of pneumonia. Pediatr Infect Dis J 2000;19:373-377.McMillan JA, Sandström C, Weiner LB, Forbes BA, Woods M, Howard T, Poe L, Keller K,

Corwin RM, Winkelman JW. Viral and bacterial organisms associated with acute pharyngitisin a school-aged population. J Pediatr 1986;109:747-752.

Mellencamp M, Preheim L, McDonlad T. Isolation and characterization of circulating immunecomplexes for patients with pneumococcal pneumonia. Infect Immun 1987;55:1737-1742.

Mimica I, Donoso E, Howard JE, Lederman GW. Lung puncture in the etiological diagnosis ofpneumonia: a study of 543 infants and children. Am J Dis Child 1971;122:278-282.

Murphy TF, Henderson FW, Clyde WA Jr, Collier AM, Denny FW. Pneumonia: An eleven-year studyin a pediatric practise. Am J Epidemiol 1981;113:12-21.

Musher DM. Infections caused by Streptococcus pneumoniae: clinical spectrum, pathogenesis,immunity, and treatment. Clin Infect Dis 1992;14:801-807.

Musher DM, Mediwala R, Phan HM, Chen G, Baughn RE. Nonspecificity of assaying for IgGantibody to pneumolysin in circulating immune complexes as a means to diagnosepneumococcal pneumonia. Clin Infect Dis 2001;32:534-538.

Mäkelä M, Puhakka T, Ruuskanen O, Leinonen M, Saikku P, Kimpimäki M, Blomqvist S, Hyypiä T,Arstila P. Viruses and bacteria in the etiology of the common cold. J Clin Microbiol1998;36:3155-3159.

Mäkelä PH, Karma P, Sipilä M, Pukander J, Leinonen M. Possibilities of preventing otitis media byvaccination. Scand J Infect Dis 1983;39:34-38.

Nohynek H, Eskola J, Laine E, Halonen P, Ruutu P, Saikku P, Kleemola M, Leinonen M. The causesof hospital-treated acute lower respiratory tract infection in children. Am J Dis Child1991;145:618-622.

84

Nohynek H, Eskola J, Kleemola M, Jalonen E, Saikku P, Leinonen M. Bacterial antibody assays inthe diagnosis of acute lower respiratory tract infection in children. Pediatr Infect Dis J1995;14:478-484.

Normann E, Gnarpe J, Gnarpe H, Wettergren B. Chlamydia pneumoniae in children attendingday-care centers in Gävle, Sweden. Pediatr Infect Dis J 1998;17:474-478.

Norrby SR. Short-term treatment of uncomplicated lower urinary tract infections in women.Rev Infect Dis 1990;12:458-466.

O’Brien KL, Walters MI, Sellman J, Quinlisk P, Regnery H, Schwartz B, Dowell SF. Severepneumococcal pneumonia in previously healthy children: the role of preceding influenzainfection. Clin Infect Dis 2000;30:784-789.

O’Neill KP, Lloyd-Evans N, Campbell H, Forgie IM, Sabally S, Greenwood BM. Latex agglutinationtest for diagnosing pneumococcal pneumonia in children in developing countries. Brit Med J1989;298:1061-1064.

Ozanne G, Lefebvre J. Specificity of microimmunofluorescence assay for the serodiagnosis ofChlamydia pneumoniae infection. Cand J Microbiol 1992;38:1185-1189.

Paisley JW, Lauer BA, McIntosh K, Glode MP, Schachter J, Rumack C. Pathogens associated withacute lower respiratory tract infection in young children. Pediatr Infect Dis J 1984;3:14-19.

Pallares R, Liñares J, Vadillo M, Cabellos C, Manresa F, Viladrich PF, Martin R, Gudiol F.Resistance to penicillin and cephalosporin and mortality from severe pneumococcalpneumonia in Barcelona, Spain. N Engl J Med 1995;333:474-480.

Paton AW, Paton JC, Lawrence AJ, Goldwater PN, Harris RJ. Rapid detection of respiratorysyncytial virus in nasopharyngeal aspirates by reverse transcription and polymerase chainreaction amplification. J Clin Microbiol 1992;30:901-904.

Paton JC, Adrew PW, Boulnois GJ, Mitchell TJ. Molecular analysis of the pathogenicity ofStreptococcus pneumoniae: The role of pneumococcal proteins. Ann Rev Microbiol1993;47:89-115.

Pechère JC (ed.). Community-acquired pneumonia in children, 1st ed., p. 7-12. Cambridge MedicalPublications, U.K., 1995.

Peltola H. C-reactive protein for rapid monitoring of infections of the central nervous system.Lancet 1982;1:980-983.

Peltola H, Räsänen JA. Quantitative C-reactive protein in relation to erythrocyte sedimentation rate,fever, and duration of antimicrobial therapy in bacteraemic diseases of childhood.J Infect 1982;5:257-267.

Peltola H. C-reactive protein in rapid differentiation of acute epiglottitis from spasmodic croup andacute laryngotracheititis: a preliminary report. J Pediatr 1983;102:713-715.

Peltola H, Laipio ML, Siimes MA. Quantitative C-reactive protein (CRP) determined by animmunoturbidimetric method in rapid differential diagnosis of acute bacterial and viraldiseases of children. Acta Paediatr Scand 1984;73:273-274.

Peltola H, Jaakkola M. C-reactive protein in early detection of bacteremic versus viral infection inimmunocompetent and compromised children. J Pediatr 1988;113:641-646.

Peltola H, Anttila M, Renkonen O-V, the Finnish study Group. Randomised comparison ofchloramphenicol, ampicillin, cefotaxime, and ceftriaxone for childhood bacterial meningitis.Lancet 1989;i:1281-1287.

Peltola H, Unkila-Kallio L, Kallio MJT. Simplified treatment of acute staphylococcal osteomyelitisof childhood. Pediatrics 1997;99:846-850.

Peltola H. Worldwide Haemophilus influenzae type b disease at the beginning of the 21st century:global analysis of the disease burden 25 years after the use of the polysaccharide vaccine anda decade after the advent of conjugates. Clin Microbiol Rev 2000;13:302-317.

Perlmutt LM, Braun SD, Newman GE, Oke EJ, Dunnick NR. Timing of chest film follow-up aftertransthoracic needle aspiration. AJR Am J Roentgenolog 1986;146:1049-1050.

Pichichero ME, Disney FA, Talpey WB, Green JL, Francis AB, Roghmann WB, Hoekelman RA.Adverse and beneficial effects of immediate treatment of group A betahemolyticstreptococcal pharyngitis with penicillin. Pediatr Infect Dis J 1987;6:635-643.

Pichichero ME, Cohen R. Shortened course of antibiotic therapy for acute otitis media, sinusitis andtonsillopharyngitis. Pediatr Infect Dis J 1997;16:680-695.

85

Pines A, Raafat H, Khorasani M, Mullinger BM. Cefuroxime and ampicillin compared in thetreatment of lower respiratory tract infections. Chemother 1981;27:459-465.

Pírez M, Martínez O, Ferrari AM, Nairac A, Montano A, Rubio I, Sarácha MJ, Brea S, Picon T,Pinchack MC, Torello P, Algorta G, Mogdasy MC. Standard case management of pneumoniain hospitalized children in Uruquay, 1997 to 1998. Pediatr Infect Dis J 2001;20:283-289.

Pradier C, Dunais B, Carsenti-Etesse H, Dellamonica P. Pneumococcal resistance patterns in Europe.Eur J Clin Microb Infect Dis 1997;16:644-647.

Putto A, Ruuskanen O, Meurman O, Ekblad H, Korvenranta H, Mertsola J, Peltola H, Sarkkinen H,Viljanen MK, Halonen P. C-reactive protein in the evaluation of febrile illness. Arch DisChild 1986;61:24-29.

Putto-Laurila A, Mertsola J, Ruuskanen O. Viral causes of tonsillitis and fever unresponsive toantibiotic therapy. Pediatr Infect Dis J 1999;18:71-72.

Ramsey BW, Marcuse EK, Foy HM, Cooney MK, Allan I, Brewer D, Smith AL. Use of bacterialantigen detection in the diagnosis of pediatric lower respiratory tract infections. Pediatrics1986;78:1-9.

Rapola S, Jäntti V, Haikala R, Syrjänen R, Carlone GM, Sampson JS, Briles DE, Paton JC,Takala AK, Kilpi TM, Käyhty H. Natural development of antibodies to pneumococcal surfaceprotein A, pneumococcal surface adhesin A, and pneumolysin in relation to pneumococcalcarriage and acute otitis media. J Infect Dis 2000;182:1146-1152.

Rapola S, Kilpi T, Lahdenkari M, Mäkelä PH, Käyhty H. Antibody response to the pneumococcalproteins pneumococcal surface adhesin A and pneumolysin in children with acute otitismedia. Pediatr Infect Dis J 2001;20:482-487.

Rayner MG, Zhang Y, Gorry MC, Chen Y, Post JC, Ehrlich GD. Evidence of bacterial metabolicactivity in culture-negative otitis media with effusion. JAMA 1998;279:296-299.

Riley I, Carrad E, Gratten H, Gratten M, Lovuru K, Phillips P, Pratt D, Rose A, Siwi H, Smith D.The status of research on acute respiratory infections in children in Papua New Guinea.Pediatr Res 1983;17:1041-1043.

Roord JJ, Wolf BH, Gossens MM, Kimpen JL. Prospective open randomized study comparingefficacies and safeties of a 3-day course of azithromycin and a 10-day course of erythromycinin children with community-acquired acute lower respiratory tract infections. AntimicrobAgents Chemother 1996;40:2765-2768.

Rosenfeld RM, Verteres JE, Carr J, Cipolle RJ, Uden DL, Giebink GS, Canafax DM. Clinicalefficacy of antimicrobial drugs for acute otitis media: meta-analysis of 5000 children fromthirty-three randomized trials. J Pediatr 1994;128:355-367.

Rubin LR. Pneumococcal vaccine. Ped Clin North Am 2000;47:269-285.Rudolph KM, Parkinson AJ, Black CM, Mayer LW. Evaluation of polymerase chain reaction for

diagnosis of pneumococcal pneumonia. J Clin Microbiol 1993;31:2661-2666.Ruiz-Gonzáles A, Nogues A, Falguera M, Porcel JM, Huelin E, Rubio-Caballero M. Rapid detection

of pneumococcal antigen in lung aspirates: comparison with culture and PCR technique.Respir Med 1997;91:201-206.

Ruiz-Gonzáles A, Falguera M, Nogues A, Rubio-Caballero M. Is Streptococcus pneumoniae theleading cause of pneumonia of unknown etiology? A microbiologic study of lung aspirates inconsecutive patients with community-acquired pneumonia. Am J Med 1999;106:385-390.

Ruuskanen O, Putto A, Sarkkinen H, Meurman O, Irjala K. C-reactive protein in respiratory virusinfections. J Pediatr 1985;107:97-100.

Ruuskanen O, Arola M, Putto-Laurila A, Mertsola J, Meurman O, Viljanen MK, Halonen P. Acuteotitis media and respiratory virus infections. Pediatr Infect Dis J 1989;8:94-99.

Ruuskanen O, Nohynek H, Ziegler T, Capeding R, Rikalainen H, Huovinen P, Leinonen M.Pneumonia in childhood: Etiology and response to antimicrobial therapy. Eur J ClinMicrobiol Infect Dis 1992;11:217-223.

Ruuskanen O, Merstola J. Childhood community-acquired pneumonia. Semin Resp Infect1999;14:163-172.

Salo P, Örqvist Å, Leinonen M. Diagnosis of bacteremic pneumococcal pneumonia by amplificationof pneumolysin genefragment in serum. J Infect Dis 1995;171:479-482.

86

Scott JAG, Hall AJ. The value and complications of percutaneous transthoracic lung aspiration forthe etiologic diagnosis of community-acquired pneumonia. Chest 1999;116:1716-1732.

Scott JAG, Hall AJ, Muyodi C, Lowe B, Ross M, Chohan B, Mandaliya K, Getambu E, Gleeson F,Dropniewski F, Marsh K. Aetiology, outcome, and risk factors for mortality among adultswith acute pneumonia in Kenya. Lancet 2000;355:1225-1230.

Schmidt HJ, Fink RJ. Rhinovirus as a lower respiratory tract pathogen in infants. Pediatr InfectDis J 1991;10:700-702.

Seppälä H, Klaukka T, Vuopio-Varkila J, Muotiala A, Helenius H, Lager K, Huovinen P. The effectof changes in the consumption of macrolide antibiotics on erythromycin resistance amonggroup A streptococci in Finland. N Engl J Med 1997;337:441-446.

Shann F, Gratten M, Germer S, Linnemann V, Hazlett D, Payne R. Aetiology of pneumonia inchildren in Goroka Hospital, Papua New Guinea. Lancet 1984;2:537-541.

Shann F. Etiology of severe pneumonia in children in developing countries. Pediatr InfectDis J 1986;5:247-252.

Shann F, Linnemann V, Gratten M. Serum concentrations of penicillin after intramuscularadministration of procaine, benzyl, and benethamine penicillin in children with pneumonia.J Pediatr 1987;110:299-302.

Shann F. The management of pneumonia in children in developing countries. Clin Infect Dis1995;21:S218-S225.

Shinefield H, Black S. Efficacy of pneumococcal conjugate vaccines in large scale field trials.Pediatr Infect Dis J 2000;19:394-397.

Sidal M, Oguz F, Unuvar A, Sarbat G, Neyzi O. Trial of co-trimazole versus procaine penicillin Gand bentzathin penicillin + procaine penicillin G in the treatment of childhood pneumonia.J Trop Pediatr 1994;40:301-304.

Silverman M, Stratton D, Diallo A, Egler LJ. Diagnosis of acute bacterial pneumonia in Nigerianchildren. Value of needle aspiration of lung and of countercurrent immunoelectrophoresis.Arch Dis Child 1977;52:925-931.

Sinha DP, Hughes JR. Lung tap in lower respiratory tract infections in children. Indian Pediatr1966;3:385-392.

Sniadack DH, Schwartz B, Lipman H, Bogaerts J, Butler JC, Dagan R, Echaniz-Aviles G,Lloyd-Evans N, Fenoll A, Girgis NI, Henrichsen J, Klugman K, Lehmann D, Takala AK,Vandepitte J, Gove S, Breiman RF. Potential interventions for the prevention of childhoodpneumonia: geographic and temporal differences in serotype and serogroup distribution ofsterile site pneumococcal isolates from children – implications for vaccine strategies.Pediatr Infect Dis J 1995;14:503-510.

Sormunen P, Kallio MJ, Kilpi T, Peltola H. C-reactive protein is useful in distinguishing Gram stain-negative bacterial meningitis from viral meningitis in children. J Pediatr 1999;134:725-729.

Straus WL, Quazi SA, Kundi Z, Nomani NK, Schwartz B, the Pakistan Co-trioxazole Study Group.Antimicrobial resistance and clinical effectiveness of co-trimoxazole versus amoxycillinfor pneumonia among children in Pakistan: randomised controlled trial.Lancet 1998;352:270-274.

Swartz MN. Use of antimicrobial agents and drug resistance. N Engl J Med 1997;337:491-492.Swingler GH. Radiologic differentiation between bacterial and viral lower respiratory infection in

children: a systemic literature review. Clin Pediatr 2000;39:627-633.Tan TQ, Mason Jr EQ, Barson WJ, Wald ER, Schutze GE, Bradley J, Arditi M, Givner LB, Yogev R,

Kim KS, Kaplan SL. Clinical characteristics and outcome of children with pneumoniaattributable to penicillin-susceptible and penicillin non-susceptible Streptococcuspneumoniae. Pediatrics 1998;102:1369-1375.

Thomas HM, Parker F. Results of antemortem lung punctures in lobar pneumonia: their bearing onthe mechanism of crisis. Arch Intern Med 1920;26:125-132.

Thornsberry C, Ogilvie PT, Holley HP, Sahm DF. Survey of susceptibilities of Streptococcuspneumoniae, Haemophilus influenzae, and Moraxella catarrhalis isolates to 26 antimicrobialagents: a prospective U.S. study. Antimicrob Agents Chemother 1999;43:2612-2623.

Toikka P, Nikkari S, Ruuskanen O, Leinonen M, Mertsola J. Pneumolysin-PCR based diagnosis ofinvasive pneumococcal infection in children. J Clin Microbiol 1999a;37:633-637.

87

Toikka P, Virkki R, Mertsola J, Ashorn P, Eskola J, Ruuskanen O. Bacteremic pneumococcalpneumonia in children. Clin Infect Dis 1999b;29:568-572.

Toikka P, Juvén T, Virkki R, Leinonen M, Mertsola J, Ruuskanen O. Streptococcus pneumoniae andMycoplasma pneumoniae coinfection in community-acquired pneumonia. Arch Dis Child2000;83:413-414.

Triga MG, Syrogiannopoulos GA, Thoma KD, Fezoulidis IB, Pastromas VG, Beratis NG. Correlationof leucocyte count and erythrocyte sedimentation rate with the day of illness in presumedbacterial pneumonia of childhood. J Infect 1998;36:63-66.

Tsarouhas N, Shaw KN, Hodinka RL, Bell LM. Effectiveness of intramuscular penicillin versus oralamoxicillin in the early treatment of outpatient pediatric pneumonia. Ped Emerg Care1998;14:338-341.

Tuomanen E, Rich R, Zak O. Induction of pulmonary inflammation by components of thepneumococci cell surface. Am Rev Respir Dis 1987;135:869-874.

Turner RB, Lande AE, Chase P, Hilton N, Weinberg D. Pneumonia in pediatric outpatients:Cause and clinical manifestations. J Pediatr 1987;111:194-200.

Tuuminen T, Palomäki P, Paavonne J. The use of serologic tests for the diagnosis of chlamydialinfections. J Microbiol Meth 2000;42:265-279.

Ueyama T, Kurono Y, Shirabe K, Takeshita M, Mogi G. High incidence of Haemophilus influenzaein nasopharyngeal secretions and middle ear effusions as detected by PCR. J Clin Microbiol1995;33:1835-1838.

Ukkonen P, Hovi T, von Bonsdorff C, Saikku P, Penttinen K. Age-specific prevalence of complement- fixing antibodies to sixteen viral antigens: A computer analysis of 58,500 patients coveringa period of eight years. J Med Virol 1984;13:131-148.

Ukkonen P, Julkunen I. Preparation of nasopharyngeal secretions for immuno-fluorescence byone-step centrifugation through Percoll. J Virol Meth 1987;15:291-301.

Unkila-Kallio L, Kallio MJ, Eskola J, Peltola H. Serum C-reactive protein, erythrocytesedimentation rate, and white blood cell count in acute hematogenous osteomyelitis ofchildren. Pediatrics 1994;93:53-62.

Vaahtera M, Kulmala T, Maleta K, Cullinan T, Salin ML, Ashorn P. Epidemiology and predictors ofinfant morbidity in rural Malawi. Pediatr Perinat Epidemiol 2000;14:363-371.

Vesanen M, Piiparinen H, Kallio M, Vaheri A. Detection of herpes simplex virus DNA incerebrospinal fluid samples using the polymerase chain reaction and microplatehybridization. J Virol Meth 1996;59:1-11.

Virolainen A, Jero J, Chattophadhay P, Karma P, Eskola J, Leinonen M. Comparison of serumantibodies to pneumolysin with those to pneumococcal capsular polysaccharides in childrenwith acute otitis media. Pediatr Infect Dis J 1996;15:128-133.

Vogel F. Sequential therapy in the hospital management of lower respiratory infections. Am J Med1995;99:S14-S19.

Vuori-Holopainen E, Peltola H. Reappraisal of lung tap: review of an old method for better etiologicdiagnosis of childhood pneumonia. Clin Infect Dis 2001;32:715-726.

Wall RA, Corrah PT, Mabey DCW, Greenwood BM. The etiology of lobar pneumonia inThe Gambia. Bull WHO 1986;64:553-558.

Wang D, Clemetns P, Lauwers S. Comparison of bacterial culture results in bronchoalveolar lavageand nasal lavage fluid in children with pulmonary infection. Int J Pediatr Otorhinolaryngol1994;28:149-155.

Waris ME, Toikka P, Saarinen T, Nikkari S, Meurman O, Vainionpää R, Mertsola J, Ruuskanen O.Diagnosis of Mycoplasma pneumoniae pneumonia in children. J Clin Microbiol1998;36:3155-3159.

Wendelinen Dorigo-Zetsma J, Wilbrink B, van der Nat H, Bartelds AIM, Heijnen MA, Dankert J.Results of molecular detection of Mycoplasma pneumoniae among patients with acuterespiratory infection and in their household contacts reveals children as human reservoirs.J Infect Dis 2001;183:675-678.

Wesley AG, Sutton JB, Widrich AJ. The aetiology of pneumonia associated with measles in Bantuchildren. S Afr Med J 1971;45:1402-1404.

88

World Health Organization. Acute respiratory infections in children: Case management in smallhospitals in developing countries. A manual for doctors and other senior health workers.Geneva: World Health organization, 1994, (WHO/ARI/90.5).

Wubbel L, Muniz L, Ahmed A, Trujillo M, Carubelli C, Mccoig C, Abramo T, Leinonen M,McCracken GH. Etiology and treatment of community-acquired pneumonia in ambulatorychildren. Pediatr Infect Dis J 1999;18:98-104.

Yang Y, Shen X, Vuori-Holopainen E, Leboulleux D, Wang YJ, Leinonen M, Hedman K,Linnavuori K, Peltola H. Seroetiology of acute lower respiratory infections amonghospitalized children in Beijing. Pediatr Infect Dis J 2001;20:52-58.

Zalacain R, Llorente JL, Gaztelurrutia L, Pijoan JI, Sobradillo V. Influence of three factors on thediagnostic effectiveness of transthoracic needle aspiration in pneumonia.Chest 1995;107:96-100.

Zhanel GG, Karlowsky JA, Low DE, The Canadian Respiratory Infection Study Group, Hoban DJ.Antibiotic resistance in respiratory tract isolates of Haemophilus influenzae and Moraxellacatarrhalis collected from across Canada in 1997-1998. J Antimicrob Chemother2000;45:655-662.