Eur J Clin Microbiol Infect Dis (2004) 23: 677–681DOI 10.1007/s10096-004-1197-2

ARTICLE

A. Pérez . P. Sala . M. Giménez . M. Sierra . A. Esteve .A. Alonso . M. Quesada . F. Raspall . V. Ausina .C. Rodrigo

Pneumococcal bacteremia in children: an 8-year review in twohospitals in Barcelona

Published online: 18 August 2004# Springer-Verlag 2004

Abstract In this study, 90 episodes of pneumococcalbacteremia that occurred over an 8-year period in twohospitals in Barcelona were analyzed retrospectively todetermine the clinical and bacteriological characteristics ofpneumococcal bacteremia, the risk factors for antibioticresistance, the outcome, and the vaccine coverage. Themean age of the patients was 3.1 years and the male/female ratio was 1.7. The overall rates of penicillin-non-susceptible, cefotaxime-non-susceptible, and erythromy-cin-resistant isolates were 48.8, 24.4, and 25.5%, respec-tively. Antibiotic resistance was associated with childrenunder the age of 2 years and with previous antibiotictreatment. The percentage of antibiotic resistance washigher in the nine episodes that occurred in patients withan underlying illness. The most prevalent serotypesidentified were 1, 14, 6B, 18C, 5, and 19A. Serotypes6A/B, 14, and 19A/F were isolated primarily fromchildren under 2, whereas serotypes 1 and 5 were

recovered more frequently from older children. Apparentrelationships between serotypes 6A/B, 14, and 19A/F andoccult bacteremia and between serotypes 1 and 5 andbacteremic pneumonia were confounded by the agevariable. The proportion of bacteremic episodes preven-table by all (7-valent, 9-valent, and 11-valent) of theconjugate pneumococcal vaccines was 60% in childrenunder 2. In older children, the serotype coverage rate forthe three formulations was 48, 87, and 87%, respectively.In summary, these data expand upon previous Spanishstudies in which serotypes 1 and 5 were reported to beamong the leading causes of severe systemic pneumococ-cal infections in children over 2, findings that should betaken into consideration when planning vaccine pro-grammes.

Introduction

Streptococcus pneumoniae is the leading cause of com-munity-acquired bacteremia and invasive bacterial infec-tions in children worldwide [1, 2]. Moreover, the increasein antibiotic resistance in invasive pneumococcal diseasehas become a major public health concern throughout theworld, since the approach to treatment has been renderedcomplicated, especially in the case of meningitis [1].

The recent advent of 7-valent pneumococcal conjugatevaccine has given us the opportunity to reduce the burdenof pneumococcal infections in our pediatric population.Therefore, the Advisory Committee on Vaccines of theSpanish Association of Pediatrics has recommended itsuse in healthy children under 2 and in older children withrisk factors for invasive pneumococcal disease [3].Although the heptavalent vaccine was licensed in ourcountry in June 2001, to date, it has not been introducedwithin the routine immunization programmes of anyautonomous region in Spain. Moreover, for reasons notentirely understood, geographical variations in incidencerates, clinical characteristics, serotype distribution, andpenicillin resistance prevalence in invasive pneumococcaldisease have been observed among several developed

A. Pérez (*) . M. Giménez . M. Quesada . V. AusinaServicio de Microbiología, Hospital Germans Trias i Pujol,Carretera del Canyet s/n, Badalona,08916 Barcelona, Spaine-mail: andres@ns.hugtip.scs.esTel.: +34-93-4978894Fax: +34-93-4978895

A. Alonso . C. RodrigoServicio de Pediatría, Hospital Germans Trias i Pujol,Carretera del Canyet s/n, Badalona,08916 Barcelona, Spain

P. Sala . F. RaspallServicio de Pediatría, SCIAS-Hospital de Barcelona,Avenida Diagonal 660,08034 Barcelona, Spain

M. SierraServicio de Microbiología, SCIAS-Hospital de Barcelona,Avenida Diagonal 660,08034 Barcelona, Spain

A. EsteveCESCAT, Hospital Germans Trias i Pujol,Carretera del Canyet s/n, Badalona,08916 Barcelona, Spain

countries [4, 5]. Hence, exhaustive knowledge of thecharacteristics of invasive pneumococcal disease in ourregion is of special importance before immunization withpneumococcal conjugate vaccines is introduced.

In this context, a retrospective study of pneumococcalbacteremia occurring in children over an 8-year period intwo hospitals in Barcelona was conducted to determine therelationship between bacteremia and age of the patient,source, antibiotic susceptibility of the causative isolate,risk factors for antibiotic resistance, outcome, serotype,and likelihood of vaccine coverage.

Materials and methods

Setting

The Germans Trias i Pujol (HUGTiP) is a 573-beduniversity teaching hospital affiliated with the UniversidadAutónoma de Barcelona that provides healthcare for apopulation of 686,987 inhabitants. The pediatric depart-ment has 38 beds and approximately 2,000 admissions and25,000 visits to the pediatric emergency department peryear. SCIAS-Hospital de Barcelona (SHB) is a 300-bednon-teaching tertiary medical center that serves anestimated population of 170,000 inhabitants. The pediatricdepartment has 16 beds with approximately 750 admis-sions and 7,000 emergency room visits per year.

Clinical methods

A “case” was defined as a child under the age of 15 yearswith a positive blood culture for S. pneumoniae in thepediatric departments of HUGTiP and SHB from 1January 1994 to 31 December 2001. “Occult bacteremia”was defined as pneumococcal bacteremia in well-appear-ing children aged 3–36 months with fever and without aclear focus of infection, whereas “primary bacteremia”was defined as pneumococcal bacteremia with no obvioussource in patients older than 3 years. Patients’ charts werereviewed for demographic characteristics and clinicalfeatures, including source of bacteremia, underlyingillness, risk factors for antibiotic resistance, laboratorytests (leukocyte, neutrophil, and band-form counts), lengthof hospital stay, and outcome.

At both centers, blood samples for culture are routinelyobtained from children aged 3–36 months who present tothe emergency department with fever over 39°C and noobvious focus of infection (acute otitis media by itself isnot considered a focus). Similarly, blood cultures areperformed for all febrile patients under 3 months of age,for febrile children who are immunocompromised or arereceiving immunosuppressive treatment, and for childrenwith toxic appearance or evidence of severe bacterialinfection. None of the patients received the 7-valentconjugate vaccine.

Microbiologic methods

Blood samples were processed in continuous-monitoringblood culture systems. At HUGTiP, the Vital blood culturesystem (bioMérieux, Marcy-L’Etoile, France) was usedfrom 1994 to 1996, whereas the BacT/Alert system(Organon Teknika, Durham, NC, USA) has been usedsince 1997. At SHB, specimens were inoculated intobottles of the Vital blood culture system.

All pneumococcal isolates were presumptively identi-fied by alpha-hemolysis on sheep blood agar andconfirmed by optochin susceptibility testing (Bio-Rad,Marnes-la-Coquette, France) and an agglutination test(Slidex Pneumo-Kit; bioMérieux). Serotyping was carriedout at the Laboratorio de Referencia de Neumococos(Instituto de Salud Carlos III, Majadahonda, Madrid) byQuellung reaction and/or dot blot assay (Statens Serumin-stitut, Copenhagen, Denmark). Minimum inhibitory con-centrations (MICs) were determined at HUGTiP bymicrodilution with sheep-blood-supplemented Mueller-Hinton broth (Sensititre; Trek Diagnostic System, EastGrinstead, UK) and at SHB using E-test strips (ABBiodisk, Solna, Sweden).

Resistance to penicillin, cefotaxime, and erythromycinwas defined according to the criteria of the NationalCommittee for Clinical Laboratory Standards. In ourstudy, all isolates for which the MIC of penicillin was≥0.1 μg/ml were considered penicillin non-susceptible,and all isolates for which the MIC of cefotaxime was≥1 μg/ml were considered cefotaxime non-susceptible.

Statistical analysis

The association between several characteristics wasassessed by the chi-square test with Yates’ correction forcontinuity or Fisher’s exact test when an expected valuewas less than 5. Additionally, the analysis of risk factorsfor bacteremia due to antibiotic-resistant strains wasenhanced by the determination of odds ratios (ORs) andtheir 95% confidence intervals (95%CIs). The t-test wasused to compare the mean value of the main epidemio-logical and biological characteristics between penicillin-susceptible and penicillin-non-susceptible isolates frombacteremic episodes with a clinical source other thanmeningitis. A logarithm in base 10 transformation wasapplied for variables presenting non-homogeneity ofvariance (Levene’s test). For all statistical analyses, alevel of 0.05 was considered significant.

To assess the relationship between serotypes 6A/B, 14,and 19A/F and occult bacteremia and also betweenserotypes 1 and 5 and bacteremic pneumonia, multiplecomparisons were performed using a Bonferroni-adjustedlevel of significance (i.e., a level of 0.025 was consideredsignificant). In order to control the potentially confound-ing influence of the age variable, the analysis was adjustedusing the Mantel-Haenszel test. Haldane’s modificationwas used when any observed value was equal to 0 byadding 1 to every cell of the frequency table [6].

678

Results

Ninety episodes of pneumococcal bacteremia werediagnosed over the 8-year period. The demographicfeatures of the patients and the foci of bacteremic episodesare shown in Table 1. Nine episodes were observed inpatients with pre-existing medical conditions: acute lym-phoblastic leukemia (n=1), septo-optic dysplasia (n=1),chickenpox (n=1), tetralogy of Fallot (n=1), post-traumaticcerebrospinal fluid fistula (n=1), Di George’s syndrome(n=1), Hodgkin’s lymphoma (n=1), and ethmoid enceph-alocele (n=2 episodes in 1 child).

The overall rates of penicillin-non-susceptible, cefotax-ime-non-susceptible, and erythromycin-resistant isolateswere 48.8, 24.4, and 25.5%, respectively (Table 1).Attendance at a day-care center could be determined in atotal of 41 of 57 patients. Among these children, nosignificant relation between penicillin-non-susceptibleisolates and care outside of the home was found(OR=2.59, P>0.2).

Bacteremia due to penicillin-non-susceptible, cefotax-ime-non-susceptible, and erythromycin-resistant pneumo-cocci was observed more frequently in patients under 2years of age (OR=4.21, 95%CI 1.73–10.25, P=0.002;

OR=3.08, 95%CI 1.14–8.35, P=0.02; OR=4.19, 95%CI1.51–11.60, P=0.004; respectively). A strong associationwas observed between consumption of β-lactam anti-biotics and macrolides during the 7 days before theconsultation at the pediatric emergency department andrecovery of penicillin-non-susceptible isolates (OR=11.85,95%CI 2.49–56.31, P=0.001) and erythromycin-resistantisolates (OR=10.33, 95%CI 2.37–45.01, P=0.001). Theprevalence of resistance to cefotaxime and erythromycinwas highly significant among penicillin-non-susceptibleisolates (P<0.0001). Among the patients with underlyingillness, the percentage of penicillin-non-susceptible iso-lates, cefotaxime-non-susceptible isolates, and erythromy-cin-resistant isolates was 66.6, 22.2, and 33.3%, respec-tively.

Among episodes with a clinical source other thanmeningitis, no significant differences were found betweenbacteremia caused by penicillin-non-susceptible isolatesand bacteremia caused by penicillin-susceptible isolates inthe duration of hospital stay or the leukocyte, neutrophil,or band-form counts (P>0.5 for each). No deaths wereobserved in this study.

Fifty-one isolates were serotyped. The six mostcommon serotypes were 1 and 14 (each causing 15.7%of episodes), 6B and 18C (each causing 13.7%), and 5 and19A (each causing 7.8%). Serotypes 6A/B, 14, and 19A/Fwere commonly observed in children under 2, whileserotypes 1, 5, and 18C were seen primarily in olderchildren (P=0.002). An apparent association betweencertain serotypes and two particular foci of bacteremiawas also noted: serotypes 6A/B, 14, and 19A/F wereobserved mostly in children with occult bacteremia, whileserotypes 1 and 5 predominated in patients withbacteremic pneumonia. Nevertheless, when controllingfor the confounding influence of the age variable andusing the Bonferroni adjustment, no statistically signifi-cant relationship was observed (P>0.60 and P>0.025,respectively). Seventy-one percent of isolates belonging toserotypes 6A/B, 14, 19A/F, and 23A/F were penicillinnon-susceptible and 42.8% were erythromycin resistant. Incontrast, all isolates belonging to serotypes 1, 5, and 18Cwere penicillin susceptible. The clinical characteristics ofinfections caused by isolates belonging to the differentserotypes identified are summarized in Table 2.

Serotypes included in the 7-valent (4, 6B, 9V, 14, 18C,19F, and 23F), 9-valent (plus 1 and 5), and 11-valent (plus3 and 7F) conjugate vaccines would have providedpotential protection against 60% of bacteremic episodesin children under 2. On the other hand, 48% of bacteremicepisodes would have been protected by the 7-valentformulation in patients over 2, whereas in the same agegroup, this percentage would have risen to 87% using the9-valent and 11-valent vaccines. The coverage ratesprovided by the three formulations for penicillin-non-susceptible, cefotaxime-non-susceptible, and erythromy-cin-resistant serotypes were 85.7, 91, and 91%, respec-tively.

Table 1 Main characteristics of episodes of pneumococcal bacter-emia in children

Characteristic Value

Mean (median) age of patients inyears

3.1 (2.1) [(44% were <2 years ofage)]

Age range of patients 3 months–14 yearsMale/female ratio 1.7Season in which episodesoccurred

Oct.–Mar. (67%)

Focal source of bacteremia [no. (%) of episodes]Pneumonia 39 (43.3)Occult bacteremia 33 (36.6)Meningitis 8 (8.8)Primary bacteremia 5 (5.5)Cellulitis 3 (3.3)Mastoiditis 1 (1.1)Primary peritonitis 1 (1.1)Antibiotic resistance of isolates [no. (%)]a

Penicillin-susceptible 46 (51.1)Penicillin-intermediate 25 (27.7)Penicillin-resistant 19 (21.1)Cefotaxime-susceptible 68 (75.5)Cefotaxime-intermediate 21 (23.3)Cefotaxime-resistant 1 (1.1)Erythromycin-susceptible 67 (74.4)Erythromycin-resistant 23 (25.5)aThe following MIC breakpoints were used: for penicillin, ≤0.06 μg/ml (susceptible), 0.1–1 μg/ml (intermediate), and ≥2 μg/ml(resistant); for cefotaxime, ≤0.5 μg/ml (susceptible), 1 μg/ml(intermediate) and ≥2 μg/ml (resistant); and for erythromycin,≤0.25 μg/ml (susceptible) and ≥1 μg/ml (resistant).

679

Discussion

In our study, the seasonal variations and the male/femaleratio were similar to those reported previously by otherauthors [7–9]. Nevertheless, our patients were slightlyolder compared with those in previous studies [7–10]. Thismay have been due to an underdiagnosis of occultbacteremia in children under 2 in our study for thefollowing reason: a small proportion of otherwise healthyand well-appearing toddlers, who upon the initial visit tothe pediatric emergency department within 24 h of feveronset were considered to have no source of infection, weremanaged without laboratory screening except for aurinalysis and urine culture and were treated at homewithout antibiotics. In addition, most of these childrenmight have been seen in primary medical centers or inprivate pediatric practices and treated without being sent tothe pediatric emergency department for laboratory evalua-tion.

At a glance, there seems to be an apparent relationshipbetween serotypes 6A/B, 14, and 19A/F and occultbacteremia, and between serotypes 1 and 5 and bacteremicpneumonia, but this was not borne out by statisticalanalysis. The epidemiological relationship between ser-otypes and a wide spectrum of clinical syndromes hasbeen reported in other publications [10–14], but whetherthis association reflects a tropism factor of certainserotypes for a clinical site remains unclear, since thiscould be confounded by other factors such as age orgeographical variations in blood culture practices or inantibiotic resistance [2, 4].

Our percentage of penicillin-non-susceptible and eryth-romycin-resistant pneumococcal isolates concurs withthose reported in pediatric systemic infections in otherstudies in Spain [14–16]. It is worth mentioning that allerythromycin-resistant isolates were also resistant toclindamycin (data not presented). This phenotypic resis-tance (MLSB phenotype), which is due to the methylationof the ribosomal RNA, is the most prevalent described in

erythromycin-resistant strains in Spain and in otherEuropean countries [15–17]. In addition, resistance tomacrolides and third-generation cephalosporins was ob-served more frequently in penicillin-non-susceptible iso-lates. This finding is consistent with the fact thatpneumococcal isolates resistant to β-lactam antibiotics,macrolides, and other non-β-lactam antibiotics are usuallyclustered among penicillin-non-susceptible strains [15–17]. Likewise, the percentage of penicillin-non-susceptibleand erythromycin-resistant isolates was higher amongchildren with pre-existing medical conditions. It is wellknown that underlying illness is among the leading factorsassociated with antibiotic resistance and mortality due toinvasive pneumococcal disease. These reasons emphasizethe need for giving immunization priority to this group.Unlike other studies, we noticed that the attendance ofchildren at day-care centers was not a risk factor forbacteremia due to penicillin-non-susceptible strains. Onereasonable explanation for this is that the sample size ofchildren attending day-care centers was small.

Among patients with a non-central nervous systemsource of pneumococcal bacteremia, no differences werefound in length of hospital stay, laboratory findings, ormortality when clinical outcomes of penicillin-non-sus-ceptible bacteremia were compared with outcomes ofpenicillin-susceptible bacteremia. This finding concurswith those of several studies in which no significantdifferences in clinical course or mortality were observedbetween invasive pneumococcal disease due to penicillin-non-susceptible and penicillin-susceptible strains [11, 18,19].

The serotypes recovered are among the most commonserotypes reported previously from studies of invasivedisease in Spanish children [14]. The important contribu-tion of serotypes 1 and 5 (23% of isolates serotyped) isnoteworthy. In Spain serotypes 1 and 5 are particularlycommon in severe forms of systemic pneumococcalinfections in children over 2 [14]. Moreover, in otherEuropean countries such as Germany, Denmark, the UK,

Table 2 Bacteremic episodes distributed by serotype and characterized by age of patient and focal source

Serotype (no. of episodes) Median age of patient (years) Source of bacteremia (no. of episodes)

1 (8) 5.46 pneumonia (7), primary peritonitis (1)14 (8) 1.57 occult bacteremia (6), pneumonia (2)6B (7) 1.5 occult bacteremia (3), pneumonia (2), meningitis (1), cellulitis (1)18C (7) 3.5 pneumonia (3), primary bacteremia (2), occult bacteremia (2)5 (4) 5.04 pneumonia (4)19A (4) 0.8 occult bacteremia (2), pneumonia (2)6A (3) 1 occult bacteremia (1), meningitis (1), mastoiditis (1)19F (3) 1 occult bacteremia (2), meningitis (1)23F (2) 2.5 occult bacteremia (1), pneumonia (1)Other serotypes10 (1) 0.33 occult bacteremia (1)11 (1) 5.66 meningitis (1)15B (1) 1.58 meningitis (1)23A (1) 14 meningitis (1)24 (1) 1.66 occult bacteremia (1)

680

and Wales, serotype 1 is among the leading cause ofpneumococcal bacteremia in this age group [7, 8, 20].

The majority of penicillin and erythromycin-resistantisolates in this survey belonged to serotypes 6A/B, 14,19A/F, and 23A/F, which coincides with the results of anationwide surveillance study in Spain [14]. Likewise,these serotypes have become the most prevalent strains ininvasive infections in children under 2 in our country,possibly due to clonal spread among young nasopharyn-geal carriers facilitated by selective pressure attributable tooverprescription of antibiotics. By contrast, serotypes 1, 5,and 18C, frequently isolated from normally sterile sites inolder children, are usually susceptible to penicillin andmacrolides and are rarely recovered from preschoolchildren [5, 13, 14, 21]. These features may explain thehigh percentages of resistance to these antibiotics amongchildren under 2.

In this survey, an accurate enough assessment of thepotential impact of vaccine coverage was limited by thesmall proportion of pneumococcal isolates serotyped.Nonetheless, the results show that the 7-valent pneumo-coccal conjugate vaccine would not have providedprotection against serotypes 1 and 5, which are the mostcommon serotypes seen in children over 2 with bacteremicpneumonia. As these serotypes are included in the 9-valentformulation, the use of this vaccine may have increasedthe coverage, whereas the 11-valent formulation would nothave provided any additional protection to our studypopulation.

Finally, since pneumococcal conjugate vaccines may beable to induce mucosal protection and include the mostimportant penicillin-resistant and macrolide-resistant ser-otypes, they could potentially decrease nasopharyngealcolonization and infections by antibiotic-resistant strains.In our opinion, their use should be combined with trainingprogrammes that encourage the judicious use of anti-biotics, as this is the most efficient way of lowering theprevalence of penicillin-resistant and multidrug-resistantpneumococci.

References

1. Kaplan SL, Manson EO (1998) Management of infections dueto antibiotic-resistant Streptococcus pneumoniae. Clin Micro-biol Rev 11:628–644

2. Hausdorff WP, Bryant J, Kloek C, Paradiso PR, Siber GR(2000) The contribution of specific pneumococcal serogroupsto different disease manifestations: implications for conjugatevaccine formulation and use, part II. Clin Infect Dis 30:122–140

3. Comité Asesor de Vacunas de la Asociación Española dePediatría (2002) La enfermedad neumocócica y su prevención.Vacuna neumocócica conjugada heptavalente. An Esp Pediatr56:79–90

4. Hausdorff WP, Bryant J, Paradiso PR, Siber GR (2000) Whichpneumococcal serogroups cause the most important disease:implications for conjugate vaccine formulation and use, part I.Clin Infect Dis 30:100–121

5. Hausdorff WP, Siber G, Paradiso PR (2001) Geographicaldifferences in invasive pneumococcal disease rates and sero-type frequency in young children. Lancet 357:950–952

6. Haldane JB (1956) The estimation and significance of thelogarithm of a ratio of frequencies. Ann Hum Genet 20:309–311

7. Von Kries R, Siedler A, Schmitt HJ, Reinert RR (2000)Proportion of invasive pneumococcal infections in Germanchildren preventable by pneumococcal conjugate vaccines. ClinInfect Dis 31:482–487

8. Kaltoft MS, Zeuthen N, Konradsen HB (2000) Epidemiologyof invasive pneumococcal infections in children aged 0–6 yearsin Denmark: a 19-year nationwide surveillance study. ActaPaediatr Suppl 435:3–10

9. Totapally BR, Walsh WT (1998) Pneumococcal bacteremia inchildhood: a 6-year experience in a community hospital. Chest113:1207–1214

10. Pineda V, Fontanals D, Larramona H, Domingo M, Anton J,Segura F (2002) Epidemiology of invasive Streptococcuspneumoniae infections in children in an area of Barcelona,Spain. Acta Paediatr 91:1251–1256

11. Kaplan SL, Mason EO, Wald ER, Tan TQ, Schutze GE,Bradley JS et al (2002) Six-year multicenter surveillance ofinvasive pneumococcal infections in children. Pediatr Infect DisJ 21:141–147

12. Alpern ER, Alessandrini EA, McGowan KL, Bell LM, ShawKN (2001) Serotype prevalence of occult pneumococcalbacteremia. Pediatrics 108:e23. http://www.pediatrics.org/cgi/content/full/108/2/e23

13. Enright MC, Fenoll A, Griffiths D, Spratt BG (1999) The threemajor Spanish clones of penicillin-resistant Streptococcuspneumoniae are the most common clones recovered in recentcases of meningitis in Spain. J Clin Microbiol 37:3210–3216

14. Fenoll A, Jado I, Vicioso D, Berrón S, Yuste JE, Casal J (2000)Streptococcus pneumoniae in children in Spain: 1990–1999.Acta Paediatr Suppl 435:44–50

15. Pérez-Trallero E, Fernández-Mazarrasa C, García-Rey C,Bouza E, Aguilar L, García de Lomas J, Baquero F, theSpanish Surveillance Group for Respiratory Pathogens (2001)Antimicrobial susceptibilities of 1,684 Streptococcus pneumo-niae and 2,039 Streptococcus pyogenes isolates and theirecological relationships: results of a 1-year (1998–1999)multicenter surveillance study in Spain. Antimicrob AgentsChemother 45:3334–3340

16. Oteo J, Cruchaga S, Campos J, Sáez-Nieto J, Baquero F (2003)Resistencia a antibióticos en 622 Streptococcus pneumoniaeaislados en líquido cefalorraquídeo y sangre en 33 hospitalesespañoles de la Red Europea de Vigilancia de Resistencia aAntibióticos (2000) Enferm Infecc Microbiol Clin 21:12–19

17. Schmitz FJ, Verhoef J, Fluit AC, the SENTRY participantsgroup (1999) Comparative activity of 27 antimicrobialcompounds against 698 Streptococcus pneumoniae isolatesoriginating from 20 European university hospitals. Eur J ClinMicrobiol Infect Dis 18:450–453

18. Friedland IR (1995) Comparison of the response to antimicro-bial therapy of penicillin-resistant and penicillin-susceptiblepneumococcal disease. Pediatr Infect Dis J 14:885–890

19. Silverstein M, Bachur R, Harper MB (1999) Clinical implica-tions of penicillin and ceftriaxone resistance among childrenwith pneumococcal bacteremia. Pediatr Infect Dis J 18:35–41

20. Miller E, Waight P, Efstratiou A, Brisson M, Johnson A,George R (2000) Epidemiology of invasive and other pneu-mococcal disease in children in England and Wales, 1996–1998. Acta Paediatr Suppl 435:11–16

21. Porat N, Trefler R, Dagan R (2001) Persistence of two invasiveStreptococcus pneumoniae clones of serotypes 1 and 5 incomparison to that of multiple clones of serotypes 6B and 23Famong children in southern Israel. J Clin Microbiol 39:1827–1832

681