Clinical and microbiological epidemiology of Streptococcus pneumoniae bacteremia in eight French counties

ORIGINAL ARTICLE Clinical and microbiological epidemiology of Streptococcus pneumoniae bacteremia in eight French counties J. Maugein1, D. Guillemot2, M. J. Dupont3, T. Fosse4, G. Laurans5, M. Roussel-Delvallez6, J. Thierry7, M. Vergnaud8, M. Weber9 and B. Poirier10

HoÃpital Haut-LeÂveÃque, Pessac, 2INSERM 4258, Villejuif, 3HoÃpital Jean Minjoz 1, Besancon, HoÃpital de l'Archet, Nice, 5HoÃpital Nord, Amiens, 6HoÃpital Calmette, Lille, 7HIA Desgenette, Lyon, 8CHU CoÃte de Nacre 1, Caen, 9CHU de Nancy, Nancy and 10Institut SmithKline Beecham, Nanterre, France

1 4

Objective To describe the incidence of pneumococcal bacteremia not associated with infection of the central nervous system, investigate the susceptibility of bacterial isolates to b-lactams, evaluate risk factors for antibiotic resistance, and determine factors predicting patient outcome.

Over a period of 1 year, 919 Streptococcus pneumoniae isolates were collected from 919 patients with bacteremia in eight French counties. Their clinical and microbiological features were recorded. Univariate and multivariate analyses were used to determine risk factors for penicillin-non-susceptible pneumococcal bacteremia and predictors of fatal outcome.

Methods

Of the 919 patients in the study, 27% were infected with penicillin-nonsusceptible pneumococci (PNSP): 17.8% of the isolates were intermediate to penicillin, 7.2% were resistant to penicillin, 16% were intermediate to amoxicillin, and 11% were intermediate to cefotaxime; no PNSP were resistant to either of the last two antibiotics. The most common PNSP serotypes isolated were 14 (41%) and 23 (24%). A statistically signi®cant relationship between PNSP infection and age below 5 years or above 60 years in the different counties was observed by univariate and multivariate analysis. Gender, origin of bacteremia, co-morbidity, immunode®ciency, previous hospitalization and nosocomial infection were not predisposing factors associated with PNSP. The mortality rate was 20.6%: there was no increase in mortality among patients with PNSP bacteremia. Age was the strongest risk factor for mortality, but immunode®ciency also seemed to have had an impact on mortality. Clinical outcome was more closely related to clinical conditions than to the susceptibility status of S. pneumoniae.

Results

Among cases of bacteremia, 27% were caused by PNSP, but this level varies according to the counties and the age of the patients. Infection-related mortality was high, but there was no increase related to penicillin G non-susceptibility of the infecting strain.

Conclusion

Keywords

Streptococcus pneumoniae, epidemiology, bacteremia

Accepted 21 March 2002

Clin Microbiol Infect 2003; 9: 280±288

Corresponding author and reprint requests: J. Maugein, HoÃpital Haut-LeÂveÃque, Avenue Magellan, 33604 PESSAC, France Tel: ‡33 5 56 55 64 91 Fax: ‡33 5 56 55 68 67 E-mail: [email protected]

INTRODUCTION The incidence of infection with Streptococcus pneumoniae with decreased susceptibility to penicillin (PNSP) now exceeds 40% in France [1]. Invasive S. pneumoniae infections are common, and this

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases

Maugein et al Epidemiology of Streptococcus pneumoniae bacteremia 281

increasing prevalence of PNSP could constitute a therapeutic problem in the years to come. Pneumococcal strains that are resistant to penicillin are also more likely to be resistant to other classes of antimicrobial agents. In order to further our knowledge of the distribution of resistant strains, 18 Regional Pneumococcal Observatories have been set up in France. Over a period of 1 year, from January to December 1997, eight of these observatories were enlisted in a study of bacteremia. The aim of the study was to describe those invasive pneumococcal infections not associated with infection of the central nervous system, to track their antibiotic susceptibility to penicillin, amoxicillin and cefotaxime, to evaluate risk factors for antibiotic resistance, and to determine outcome in patients in relation to antibiotic susceptibility. MATERIALS AND METHODS Surveillance of bacteremia pneumococcal infections In 1997, eight counties in the network (Aquitaine, Franche ComteÂ, Lorraine, Nord-Pas de Calais, Normandie, Picardie, CoÃte d'Azur, RhoÃnes-Alpes) took part in the study, involving microbiology laboratories in teaching hospitals, general hospitals, and private hospitals. Whenever these laboratories isolated a strain of S. pneumoniae from blood cultures, they completed a `Case Report Form' containing information on demographic data, type of infection, underlying illnesses, prior hospitalization, prior antibiotic use during the previous 3 months, hospital-acquired infections (nosocomial infections were diagnosed when the signs and symptoms of disease developed following 72 h of hospitalization for an unrelated illness), and outcome within 1 month after the diagnosis of bacteremia. More than one pneumococcal isolate of the same serotype or with an identical antibiotic pro®le from the same patient within a 30-day period was considered as a single episode. Microbiology studies In each laboratory, S. pneumoniae isolates were identi®ed from their typical colony morphology on blood agar plates, their Gram strain characteristics, their optochin-disk susceptibility, and the results of latex agglutination tests. Antibiotic susceptibility testing was performed using the disk diffusion technique in Mueller±Hinton agar supplemented

with 5% horse blood, a 5-mg oxacillin disk being used to screen penicillin-resistant strains. Strains with a zone diameter of <26 mm were considered to be non-susceptible to penicillin, and the MIC was determined by the E test (AB Biodisk, Solna, Sweden). All PNSP isolates from each laboratory were sent to the regional coordinating center for determination of the MICs for penicillin, amoxicillin and cefotaxime, using the reference agar dilution method in Mueller±Hinton agar supplemented with 5% horse blood. An inoculum was prepared by culturing 10 colonies in 5 mL of brain± heart broth incubated for 3 h at 37 8C; 104 105 CFU per spot were applied using a Steers replicator onto agar plates containing serial dilutions of antibiotics. The plates were incubated at 37 8C overnight in a CO2-enriched atmosphere. Three appropriate quality control strains (Sp53858, Sp32475, and Sp1600) provided by the Pneumococci Reference Center (MD Geslin) were included in each run. The MIC was de®ned as the lowest concentration of an antimicrobial agent which completely inhibited bacterial growth. Susceptibility categories were determined using the Comite de l'Antibiogramme de la SocieÂte FrancËaise de Microbiologie (CA-SFM) guidelines [2], where pneumococci are considered to be susceptible to penicillin when the MIC is 0.06 mg/L, to exhibit intermediate susceptibility when the MIC is 0.12±1 mg/L, and to be resistant when the MIC is 2 mg/L. For amoxicillin and cefotaxime, the breakpoints were 0.5 mg/L and >2 mg/L, respectively. The PNSP serotypes were determined by using a latex reagent prepared by bioMerieux (Marcy l'Etoile, Lyon, France) and containing antibodies speci®c to the most frequent serotypes (6, 9, 14, 15, 19, and 23). Statistical analysis Continuous variables such as age were transformed into categorical variables. We then performed univariate analysis using the Pearson w2 or the Fisher exact test, where appropriate. Missing data were included in the univariate analysis as a separate value when the number of missing values was higher than 10, in order to assess the information bias. Multivariate analyses were performed by building a logistic regression model. The determination of variables to be included in the model was based on a P-value lower than 0.1. The variables ®nally taken into account were determined using a

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

282 Clinical Microbiology and Infection, Volume 9 Number 4, April 2003 backward stepwise regression and maximizing the likelihoods of the model [3]. Ninety-®ve per cent con®dence intervals were estimated using the logit method. The Stata software package (version 6.0) was used for statistical analysis.

Table 1 Distribution of 895 S. pneumoniae strains according to susceptibility to penicillin G, amoxicillin and cefotaxime

RESULTS

Adults (n ˆ 761) Susceptible 560 (73.5) Intermediate 131 (17.2) Resistant 70 (9.2) Children (n ˆ 134) Susceptible 93 (69.4) Intermediate 28 (20.9) Resistant 13 (9.7)

During the period covered by the study, 919 cases of bacteremia were identi®ed in eight regions, based on clinical microbiological laboratory records. The strains isolated from cases of bacteremia represented 13.5% of S. pneumoniae strains isolated from all the samples collected in the regions, but this percentage differed between the counties (ranging from 7% in Lorraine to 25% in RhoÃne-Alpes). Among the 919 strains isolated, the susceptibilities of 895 S. pneumoniae strains to different blactamsÐ penicillin G, amoxicillin, and cefotaximeÐwere determined using the agar dilution method. The results are shown in Table 1. PNSP levels reached 30.6% in children and 26.4% in adults. Of the 242 PNSP strains, 83 (34%) were resistant to penicillin G, while no strains were resistant to amoxicillin and cefotaxime. The percentages of strains with intermediate resistance to these two antibiotics were 16% and 11%, respectively. Of the 242 PNSP strains, 215 were serotyped using the six serotypes available. Serotype 14 was the most common, representing 41% of strains, followed by serotypes 23 (23.7%), 9 (18.6%) and 6 (11.2%). Serotypes 19 and 15 were much less common, since they only accounted for 4% and 1.4%, respectively. However, a few differences were noted between the counties: in Franche ComteÂ, serotype 23 was the most common, while in all the other regions, there was a prevalence of

Number (%) of isolates in susceptibility category Penicillin G Amoxicillin Cefotaxime 637 (83.7) 124 (16.3) ±

676 (88.8) 85 (11.2) ±

114 (83.7) 20 (15) ±

122 (91) 12 (9) ±

serotype 14, which ranged from 56% in Aquitaine to 36.7% in the Nord county. Serotype 9 varied from 26% in RhoÃne Alpes to 4.5% in Aquitaine. Among children, serotype 23 predominated (40%), followed by serotype 14, with 34.3%. The different factors associated with the isolation of S. pneumoniae with diminished susceptibility to penicillin G are summarized in Table 2. Details on counties, age, gender and nosocomial source of infection were provided for all cases of bacteremia, but only 457 cases included information on other factors. Univariate analysis of the distribution of PNSP by counties demonstrated a signi®cant difference (P ˆ 0.05). The lowest percentages of PNSP were found in RhoÃne-Alpes (19%) and Franche Comte (22%), while the highest percentage was seen in Aquitaine (36%). With the use of multivariate analysis, this difference remained signi®cant, independent of the gender and age of patients. The age-related PNSP distribution (Figure 1) showed that the highest percentage was in children between the ages of 0 and 5 years, and the

Figure 1 Rate of penicillin resistant S. pneumoniae according to the age of patients.

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

Maugein et al Epidemiology of Streptococcus pneumoniae bacteremia 283

Table 2 Factors associated with penicillin resistance in S. pneumoniae bacteremia (n ˆ 919)

Region Rhone±Alpes (n ˆ 258) Franche Comte (n ˆ 81) Picardie (n ˆ 68) Nord (n ˆ 255) Lorraine (n ˆ 104) Normandie (n ˆ 57) Aquitaine (n ˆ 96) Age (years) 0±14 (n ˆ 137) 15±60 (n ˆ 259) > 60 (n ˆ 523) Unknown (n ˆ 0) Gender Male (n ˆ 541) Female (n ˆ 372) Unknown (n ˆ 6) Type of infection Upper respiratory infection (n ˆ 40) Lower respiratory infection (n ˆ 328) Unknown (n ˆ 473) Other (78) Institutionalized patientb No (n ˆ 299) Yes (n ˆ 57) Unknown (n ˆ 563) Co-morbidity No (n ˆ 267) Yes (n ˆ 155) Unknown (n ˆ 497) Immunodeficiency No (n ˆ 197) Yes (n ˆ 167) Unknown (n ˆ 555) Previous antibiotic use No (n ˆ 239) Yes (n ˆ 93) Unknown (n ˆ 587) Previous hospitalization No (n ˆ 231) Yes (n ˆ 168) Unknown (n ˆ 520) Nosocomial infection No (n ˆ 825) Yes (n ˆ 90) Unknown (n ˆ 4)

Univariate OR

95% Confidence interval

1 1.0 1.4 1.5 1.7 1.8 2.2

1 0.6±1.9 0.8±2.6 1.0±2.2 1.0±2.8 1.0±3.4 1.3±3.6

1 0.5 0.9

Pa

0.05

0.3±0.9 0.6±1.4 ±

0.003

1.0±1.7 ±

0.08

1.3

0.6±2.8

0.5

1.3

0.6±2.8 0.8±4.4

1.3

Adjusted OR

95% Confidence interval

P

1 1.0 1.4 1.5 1.6 1.8 2.1

0.5±1.8 0.7±2.5 1±2.2 1±2.8 0.9±3.3 1.2±3.5

0.9 0.3 0.06 0.06 0.07 0.006

0.3±0.8 0.6±1.3

0.004 0.5

0.9±1.6

0.2

1 0.5 0.9 ± 1.2 ±

1

1 1.2 1

0.8±2.7 0.7±1.4

0.4

1 1.5 1.2

1.0±2.3 0.9±1.7

0.1

1 1.1 1.3

0.7±1.8 0.9±1.9

0.4

1 1.2 1

0.7±2.0 0.7±1.4

0.7

1 1.2 1.1

0.8±1.9 0.8±1.6

0.4

1 1.1

0.7±1.8

0.6

a

Global comparison of distribution between susceptible and resistant according to the class of each variable. Restricted to patients over the age of 60 years (n ˆ 523).

b

lowest percentage in those aged between 6 and 14 years. A gradual upturn was then seen in the proportion of PNSP from 15 years, reaching approximately 30% by the age of 60. Therefore,

there was a signi®cant difference in the percentage of PNSP by age (P ˆ 0.003), and multivariate analysis demonstrated that this difference persisted regardless of the counties (P ˆ 0.004).

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

284 Clinical Microbiology and Infection, Volume 9 Number 4, April 2003 Table 3 Factors associated with death in S. pneumoniae bacteremia (n ˆ 436)

Age (years) 0±14 (n ˆ 63) 15±60 (n ˆ 126) > 60 (n ˆ 247) Unknown (n ˆ 0) Sex Male (n ˆ 242) Female (n ˆ 191) Unknown (n ˆ 3) Type of infection Upper respiratory infection (n ˆ 38) Low respiratory infection (n ˆ 288) Other (n ˆ 69) Unknown (n ˆ 41) Institutionalized patient No (n ˆ 272) Yes (n ˆ 50) Unknown (n ˆ 114) Co-morbidity No (n ˆ 245) Yes (n ˆ 133) Unknown (n ˆ 58) Immunodeficiency No (n ˆ 175) Yes (n ˆ 145) Unknown (n ˆ 116) Previous antibiotic use No (n ˆ 213) Yes (n ˆ 83) Unknown (n ˆ 140) Nosocomial infection No (n ˆ 379) Yes (n ˆ 57) Unknown (n ˆ 0) Previous hospitalization No (n ˆ 231) Yes (n ˆ 168) Unknown (n ˆ 37) PRSP No (n ˆ 303) Yes (n ˆ 133)

Univariate OR

95% Confidence interval

1 4.0 7.4

1.1±13.9 2.3±24.5

0.001

1 0.8

0.5±1.3

0.5

Pa

1

Adjusted OR

95% Confidence interval

1 2.3 5.3

0.6±8.4 1.5±18.5

0.01

1.8

0.4±8.4

0.5

3.4

0.7±16.6

0.1

1 1.4

1.0±1.9

0.05

1 1.6

0.8±3.1

0.2

P 0.2

1

4.4

1.0±18.9

0.03

6.6

1.4±32.1

1 1.4

0.6±2.8

0.4

1 1.3

0.7±2.1

0.4

1 2.0

1.2±3.4

0.01

1 1.6

0.8±2.9

0.2

1 1.9

1.0±3.6

0.03

1 1.2

0.8±1.9

0.4

1 1.0

0.6±1.7

0.9

a

Comparisons of value pattern of each factor according to mortality. PRSP, penicillin-resistant S. pneumoniae.

With respect to the underlying disease associated with bacteremia, pulmonary diseases represented 73.5% of the 446 documented cases. No signi®cant difference was seen in the distribution of PNSP as a function of the initial pathology. Among documented cases, 155 patients (36.7%) presented with underlying illness, and 167 patients (45.8%) with immunode®ciency. Univariate analysis showed no signi®cant difference in PNSP

distribution as a function of whether or not patients presented with one of these factors (P ˆ 0.1 and P ˆ 0.4, respectively). Antibiotic therapy (b-lactams or another class of antibiotics) during the 3 months prior to bacteremia was recorded in 332 cases. No signi®cant difference in PNSP levels was found between patients who had received antibiotics and those who had not. PNSP were not more frequent among

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

Maugein et al Epidemiology of Streptococcus pneumoniae bacteremia 285

bacteremic patients who were living in retirement homes or with a history of prior hospitalization. Finally, in the few patients who acquired the disease during hospitalization (9.8%), the proportion of PNSP did not differ from that in community-acquired cases. Outcome was documented in 436 cases of S. pneumoniae bacteremia; 89 patients (20.6%) died during hospitalization, and 37 patients (41.5%) died within 48 h of the blood sample being drawn. Univariate analysis of factors associated with these fatalities (Table 3) showed that mortality varied as a function of region (P ˆ 0.007), age (P ˆ 0.001), immunode®ciency (P ˆ 0.01), and nosocomial acquisition (P ˆ 0.03), but was independent of strain susceptibility to penicillin G (P ˆ 0.98). No statistically signi®cant association was found between mortality and levels of penicillin resistance. The rates of mortality were, respectively, 21.3% for strains with MICs between <0.1 mg/L, 16.7% for strains with MICs 0.1 and <1 mg/L, 25.6% for strains with MICs 1 mg/L, and 20.9% for strains with MICs 2 mg/L. By multivariate analysis, only two factors remained signi®cant predictors of outcome: age over 60 years and immunode®ciency. DISCUSSION This study demonstrates that the proportion of PNSP in cases of pneumococcal bacteremia reached 27% in 1997. In the same period, on a national level, the level of PNSP among all samples analyzed was 40% [4]. It is customary to observe a lower level of PNSP among invasive strains than among all clinical isolates from various sources. In a study of S. pneumoniae pneumonia, Bedos et al. [5] demonstrated that 22% of pneumococcal pneumonia was caused by PNSP but only 4.7% in bacteremic patients. However, it should be noted that the rate of PNSP in blood cultures in France during 1999 rose signi®cantly, since it was 36% among adults and 41% among children, approaching the levels found among all samples tested during this study, which were 39% and 50% in adults and children, respectively [1]. A comparison of the results concerning susceptibility to amoxicillin and cefotaxime with those concerning susceptibility to penicillin G is in agreement with ®ndings in the literature. Marchese et al. [6] found that 62% of PNSP strains were susceptible to amoxicillin, while Vanhoof [7] observed

that MIC values of amoxicillin and cefotaxime were lower than those seen for penicillin G, that 35% of PNSP strains were susceptible to amoxicillin and cefotaxime, and that only four strains were resistant to amoxicillin and one to cefotaxime. In a study of S. pneumoniae susceptibility to b-lactams, Corrihons et al. [8] found that among strains with intermediate resistance to penicillin G, 49% were susceptible to amoxicillin and 86% to cefotaxime, and none were resistant. Among the strains resistant to penicillin G, only 12% were resistant to amoxicillin and none to cefotaxime. We found only 11.2% and 9% of isolates to be intermediately resistant to cefotaxime, respectively, in adults and children, while Kaplan et al. [9], in a recent study in children, found 7.9% of the episodes of invasive infection to be caused by S. pneumoniae with MIC for ceftriaxone of 1 mg/L, and only 0.5% to be caused by S. pneumoniae resistant to ceftriaxone. The present ®ndings show that serotypes 14 and 23 are the most common among PNSP. McKensie et al. [10] found high levels of serotype 14, while Hausdorff et al. [11] showed that, in all age groups, serotypes 1 and then 14 and 9 were most frequently isolated from blood cultures in Europe. This difference could be explained by the fact that, during our study, only PNSP were typed, while Geslin, in his report on activities in 1997 [12], noted that, in France, serotype 1 was found in only 5.4% of all blood cultures. Our results demonstrate that patients below the age of 5 years or above the age of 60 years are at higher risk of presenting with PNSP bacteremia. This age factor has been found in most studies. In a study of children, Kaplan et al. [13] found a higher level of PNSP among those below the age of 5 years, and Metlay et al. [14] noted the same among those over the age of 65. The second risk factor for PNSP bacteremia was the county of residence, and the differences observed among countries is also true for the French counties. In our study, no other risk factor for PNSP infection was found. A range of factors has been described in the literature: for example, Nava et al. [15] found that co-morbidity was a risk factor, while Metlay et al. [14] did not note any difference, even among HIV-positive patients. The emergence of S. pneumoniae as a major nosocomial pathogenic agent has been mentioned by several authors. Garcia-Leoni et al. [16] found 26.6% of patients in whom the infection was

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

286 Clinical Microbiology and Infection, Volume 9 Number 4, April 2003 acquired during hospitalization, while we noted only 10.7% of nosocomial infections and showed that this did not constitute a risk factor for the acquisition of PNSP. In contrast, other authors [15,17] have found a signi®cant difference in the distribution of PNSP, depending on whether the infection was nosocomial or community acquired. According to several authors, antibiotic therapy within 3 months of the onset of bacteremia constitutes a risk factor for PNSP infection [13,15, 17,18]. However, in a study of children, Haylund et al. [19] found that 75% of children infected by a PNSP had received antibiotic therapy, but this was true for only 60% of those infected by a susceptible strain. This author mentioned the problems encountered during his study in collecting this type of information, a factor that could also explain why, during our study, we did not ®nd that antibiotic therapy constituted a risk factor. The mortality observed in our study population was close to 20%, ranging from 4% in children to 22% in adults over the age of 60 years. This rate is close to that described in other studies [10,20]. As for the risk of death from S. pneumoniae bacteremia, our study shows that PNSP infection is not a factor predicting excess mortality, even after the adjustment for other predictive factors. These ®ndings are in agreement with those of Pallares et al. [21]. Some data [22] suggest increased mortality or complications for strains with penicillin MICs of >2 mg/L. Such differences were not noted in our study, probably because only two patients were infected with strains that had MICs 4 mg/L. The only risk factors affecting mortality were age and immunode®ciency. The age factor was found in all studies [23±25], with some authors also mentioning immunode®ciency or co-morbidity [26,27]. The present study has some limitations. First, we lacked information about the risk factors and the treatment given. Moreover, not all French counties were involved in the study, so the results cannot be generalized to a nationwide level. Finally, the study may not be exhaustive, since some data concerning pneumococcal infection might be lacking in this cohort. In conclusion, the results of this study conducted in eight French counties show that the level of PNSP among patients with bacteremia is 27%, but this may vary from region to region and according to the age of the patients. Among these PNSP, no strain was resistant to amoxicillin or cefotaxime. Infection-related mortality was high,

but there was no increase related to penicillin G non-susceptibility of the infecting strain. Age over 60 years, nosocomial acquisition and immunode®ciency were the only risk factors for mortality found in this study. LABORATORIES THAT PARTICIPATED IN REGIONAL PNEUMOCOCCAL OBSERVATORIES Aquitaine: J. Maugein, A. Armynot du Chatelet, M. C. Bezian, F. Boineau, J. P. Brochet, B. Cancet, D. Cassignard, M. P. Denjean, H. P. Doerman, B. Dutihl, Z. El-Harrif, C. Fabe, I. Fischer, S. Fourmau, J. P. Lafargue, G. Larrouy, C. Rougier, R. Sanchez, and C. Tamarelle. CoÃte d'Azur: T. Fosse, C. Buisson, Mme Raynaud, Mme Carmagnol, Mme Azencotte, M. Zumbo, M. Mora, Mme Cornelis, Mme Gabriel, M. Bensa, Mme Coppolai, Mme Langragin, Mme Chabaud, Mme Masseyef, and M. Kubinieck. Franche ComteÂ: M.-J. Dupont, B. Mulin, G. Cellier, C. Gauthier, H. Tronel, M.-E. Louvrier, L. Bruand, B. Accoceberry, P. Moritz, C. Febvre, M. Bonnin, P. Chantelat. Lorraine: M. Weber, F. Jurin, E. Deville, S. Boussard, G. Weisse, P. de Temmermann, A. Le Coustumier, P. Emerique, P. Roos, M. C. Moulhade, Y. Germain, A. Bina, M. Got, M. Urshel, and B. Aubry-Rael. Nord-Pas-de-Calais: M. Roussel-Delvallez, M. Caillaux, C. Cattoen, A. Verhaeghe, M. Bonte, S. Samaille, M. Duhamel, P. Fievet, F. Tiry, N. Reiter, D. Descamps, M. Marcolin, F. Templier, A. M. Noel, A. Varlet, and C. Morin. Normandie: M. Vergnaud, M. C. Allaire, C. Berlic, E. Bessis, A. Bouillerot, H. Bourgeois, S. Bourgeois, J. Carre-Cavelier, M. Duclos, G. Gallou, E. Godard, T. Grancher, G. Grise, J. Jehan Leang, J. F. Lemeland, C. Paris, G. Pinon, J. Poulain, C. Rennes, and H. Sep Hieng Hubert. Picardie: G. Laurans, M. Duminy, J. Heurte, C. Puzin, J. P. Darchis, P. Lemaitre, J. P. Thelier, A. Sueur, L. Dupond, J. F. Bezoc, A. Brocard, C. Bouquigny, M. T. Albertini, and M. Demange. RhoÃne Alpes: J. Thierry, M. Aubert, G. Barbe, Y. Boucaud-Maitre, M. Boyer, M. Chomarat, P. Clergeau, F. Delubac, J. L. Etienne, D. Fevre, C. Fuhrmann, B. Gravagna, M. Letouzey, A. Mandjee, P. Marthelet, R. Meley, M. E.Reverdy, A. Ros, O. Sabot, A. Sedallian, S. Tigaud, A. Tixier, and J. Tous.

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

Maugein et al Epidemiology of Streptococcus pneumoniae bacteremia 287

ACKNOWLEDGMENTS We would like to thank P. H. Thoreux for his helpful contribution to the statistical analyses. REFERENCES 1. Chomarat M, Laurans G, Chanal C et al. Evolution of resistance of clinical streptococcus pneumoniae and analysis of serotypes of PRP in France in 1999 [abstract 1794]. In: Program and abstracts of the 40th International Conference on Antimicrobial Agents and Chemotherapy, Toronto. Washington, DC: American Society for Microbiology, 2000: 114. 2. Acar J, Carret G, Cavalo JD et al. Communique 1998 du comite de l'antibiogramme de la SocieÂte FrancËaise de Microbiologie. Path Biol 1998; 46: I±XVI. 3. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, 2000. 4. Roussel-Delvallez M, Demachy MC, Vernet-Garnier V et al. ReÂsistance du pneumocoque aux antibiotiques en France en 1997. BEA 1998; 2: 183±6. 5. Bedos JP, ValleÂe E, Moine P, Geslin P, Chastang C. Pneumonies aÁ S. pneumoniae de sensibilite diminueÂe aÁ la peÂnicilline: donneÂes eÂpideÂmiologiques, facteurs de risque et impact theÂrapeutique. Med Mal Infect 1995; 25: 1±8. 6. Marchese A, Debbia E, Pesce A, Schito GC. Comparative activities of amoxicillin and 10 other oral drugs against penicillin-susceptible and resistant Streptococcus pneumoniae strains recently isolated in Italy. Clin Microbiol Infect 1998; 4: 170±3. 7. Vanhoof R. Comparative in vitro activities of amoxicillin and penicillin against Streptococcus pneumoniae isolates. J Antimicrob Chemother 1998; 42: 397±8. 8. Corrihons V, Abinars A, Arminaud du Chatelet A et al. EnqueÃte eÂpideÂmiologique reÂgionale sur la reÂsistance aux antibiotiques de Streptococcus pneumoniae; reÂsultats de l'Observatoire Pneumocoque reÂgion Aquitaine. Med Mal Infect 1997; 27(speÂcial): 16±23. 9. Kaplan SL, Masson EO, Barson WJ et al. Outcome of invasive infections outside the central nervous system caused by Streptococcus pneumoniae isolates nonsusceptible to ceftriaxone in children treated with beta-lactam antibiotics. Pediatr Infect Dis J 2001; 20: 392±6. 10. McKensie H, Reid N, Dijkhuizen RS. Clinical and microbiological epidemiology of Streptococcus pneumoniae bacteremia. J Med Microbiol 2000; 49: 361±6. 11. Hausdorff W, Bryant J, Kloek C, Paradiso PR, Siber GR. The contribution of specific pneumococcal serogroups to different disease manifestations: implications for conjugate vaccine formulation and use. Clin Infect Dis 2000; 30: 122±40.

12. Geslin P. Rapport d'activite anneÂe 1997. Paris: Centre National de ReÂfeÂrence du Pneumocoque, 1997. 13. Kaplan S, Mason EO, Barson WJ et al. Three-year multicenter surveillance of systemic pneumococcal infections in children. Pediatrics 1998; 102: 538±45. 14. Metlay JP, Hofmann J, Cetron MS et al. Impact of penicillin susceptibility on medical outcomes for adult patients with bacteriemic pneumococcal pneumonia. Clin Infect Dis 2000; 30: 520±8. 15. Nava JM, Bella F, Garau J et al. Predictive factors for invasive disease due to penicillin-resistant Streptococcus pneumoniae: a population-based study. Clin Infect Dis 1994; 19: 884±90. 16. Garcia-Leoni ME, Cercenado E, Rodeno P, Bernaldo de Quiros JCL, Martinez-Hernandez D, Bouza E. Susceptibility of Streptococcus pneumoniae to penicillin: a prospective microbiological and clinical study. Clin Infect Dis 1992; 14: 427±35. 17. Pallares R, Gudiol F, Linares J et al. Risk factors and response to antibiotic therapy in adults with bacteremic pneumoniae caused by penicillinresistant pneumococci. N Engl J Med 1987; 317: 18±22. 18. Deeks SL, Palacio R, Ruvinsky R et al. Risk factors and course of illness among children with invasive penicillin-resistant Streptococcus pneumoniae. Pediatrics 1999; 103: 409±13. 19. Haylund LA, Istre GR, Pickett DA, Welch DF, Fine DP, the Pneumococcus Study Group. Invasive pneumococcal disease in central Oklahoma: emergence of high-level penicillin resistance and multiple antibiotic resistance. J Infect Dis 1993; 168: 1532±6. 20. Ekdahl K, Martensson A, Kamme C. Bacteraemic pneumococcal infections in southern Sweden 1981± 96: trends in incidence, mortality, age-distribution, serogroups and penicillin resistance. Scand J Infect Dis 1998; 30: 257±62. 21. Pallares R, Linares J, Vadillo M et al. Resistance to penicillin and cephalosporin and mortality from severe pneumococcal pneumonia in Barcelona, Spain. N Engl J Med 1995; 333: 474±80. 22. Choi EH, Lee HJ. Clinical outcome of invasive infections by penicillin-resistant Streptococcus pneumoniae in Korean children. Clin Infect Dis 1998; 26: 1346±54. 23. Feikin DR, Schuchat A, Kolczak M et al. Mortality from invasive pneumococcal pneumonia in the era of antibiotic resistance, 1995±1997. Am J Public Health 2000; 90: 223±9. 24. Torres JM, Cardenas O, Vasquez A, Schlossberg D. Streptococcus pneumoniae bacteremia in a community hospital. Chest 1998; 113: 387±90. 25. Winston LG, Perlman JL, Rose DA, Gerberding JL. Penicillin-nonsusceptible Streptococcus pneumoniae at San Francisco general hospital. Clin Infect Dis 1999; 29: 580±5.

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288

288 Clinical Microbiology and Infection, Volume 9 Number 4, April 2003 26. Mirzanejad Y, Roman S, Talbot J, Nicolle L, the pneumococcal bacteremia study group. Pneumococcal bacteremia in two tertiary care hospitals in Winnipeg, Canada. Chest 1994; 19: 884±90.

27. Teira R, Munoz J, Zubero Z, Rojo P, Cisterna R, Santamaria JM. Epidemiologic characteristics of pneumococcal bacteremia in the era of AIDS. Enferm Infect Microbiol Clin 1992; 10: 138±42.

ß 2003 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 9, 280±288