Scholarly article on topic 'Demographic analysis of antimicrobial resistance among Streptococcus pneumoniae: worldwide results from PROTEKT 1999–2000'

Demographic analysis of antimicrobial resistance among Streptococcus pneumoniae: worldwide results from PROTEKT 1999–2000 Academic research paper on "Clinical medicine"

0
0
Share paper
OECD Field of science
Keywords
{"Antibacterial resistance" / Demography / "Respiratory tract infections" / " Streptococcus pneumoniae " / Telithromycin}

Abstract of research paper on Clinical medicine, author of scientific article — Daryl Hoban, Fernando Baquero, Vaughan Reed, David Felmingham

Summary Design: The influence of demographic factors upon antimicrobial resistance among 3362 isolates of Streptococcus pneumoniae from 25 countries was investigated, using univariate comparison and multivariate logistic regression. Results: Eleven countries had significantly higher rates (Odds ratios [OR]: 2.50–64.79) of penicillin and/or erythromycin resistance than the UK. After taking country effects into account, rates of penicillin resistance (OR 1.98) and erythromycin resistance (OR 1.89) were significantly higher among infants than adults. Fewer (OR 0.69) erythromycin-resistant isolates were collected from male than female patients. There was no difference in the incidence of penicillin or erythromycin resistance among inpatients or outpatients. Penicillin resistance was more prevalent among ear swabs than blood cultures (OR 2.07). Erm(B), the predominant macrolide resistance mechanism across all age groups, was particularly prevalent among bronchoalveolar lavage (69.1%) and sinus (68.8%) isolates. Isolates possessing both erm(B) and mef(A) were generally collected from South Korea and were most common among infants and children (10.3%) and ear samples (17.3%). Telithromycin susceptibility was >99.5%, irrespective of demography. Conclusions: Although demography had a significant impact on antimicrobial resistance of pneumococci, telithromycin remained highly active across all demographic groups.

Academic research paper on topic "Demographic analysis of antimicrobial resistance among Streptococcus pneumoniae: worldwide results from PROTEKT 1999–2000"

International Journal of Infectious Diseases (2005) 9, 262—273

ELSEVIER

Demographic analysis of antimicrobial resistance among Streptococcus pneumoniae: worldwide results from PROTEKT 1999-2000§

Daryl Hobana *, Fernando Baquerob, Vaughan Reedc, David Felminghamd

a Health Sciences Centre, Department of Clinical Microbiology, 820 Sherbrook Street, MS-673, Winnipeg, Man., Canada R3A 1R9 b Hospital Universitario Ramon y Cajal, Madrid, Spain c Micron Research Ltd, Ely, UK d GR Micro Ltd, London, UK

Received 15 December 2003; received in revised form 7 July 2004; accepted 7 July 2004 Corresponding Editor: Richard Oberhelman, New Orleans, USA

Summary

Design: The influence of demographic factors upon antimicrobial resistance among 3362 isolates of Streptococcus pneumoniae from 25 countries was investigated, using univariate comparison and multivariate logistic regression.

Results: Eleven countries had significantly higher rates (Odds ratios [OR]: 2.50— 64.79) of penicillin and/or erythromycin resistance than the UK. After taking country effects into account, rates of penicillin resistance (OR 1.98) and erythromycin resistance (OR 1.89) were significantly higher among infants than adults. Fewer (OR 0.69) erythromycin-resistant isolates were collected from male than female patients. There was no difference in the incidence of penicillin or erythromycin resistance among inpatients or outpatients. Penicillin resistance was more prevalent among ear swabs than blood cultures (OR 2.07). Erm(B), the predominant macrolide resistance mechanism across all age groups, was particularly prevalent among bronchoalveolar lavage (69.1%) and sinus (68.8%) isolates. Isolates possessing both erm(B) and mef(A) were generally collected from South Korea and were most common among infants and children (10.3%) and ear samples (17.3%). Telithromycin susceptibility was >99.5%, irrespective of demography.

Conclusions: Although demography had a significant impact on antimicrobial resistance of pneumococci, telithromycin remained highly active across all demographic groups. # 2005 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

§ This analysis was presented in part at the 10th International Congress on Infectious Disease, Singapore, 11—14 March 2002. * Corresponding author. Tel.: +1 204 787 1191; fax: +1 204 787 4699. E-mail address: dhoban@hsc.mb.ca (D. Hoban).

INTERNATIONAL SOCIETY

FOR INFECTIOUS DISEASES

http://intl.elsevierhealth.com/journals/ijid

KEYWORDS

Antibacterial

resistance;

Demography;

Respiratory tract

infections;

Streptococcus

pneumoniae;

Telithromycin

1201-9712/$30.00 © 2005 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijid.2004.07.008

Introduction

Over the last two decades, resistance to commonly used antibacterial agents among Streptococcus pneumoniae, the major bacterial pathogen implicated in community-acquired respiratory tract infections (RTIs), has spread at an alarming pace around the globe.1'2 The clinical impact of this resistance in the management of community-acquired RTIs is becoming evident, with cases of treatment failure in patients infected with a resistant pneumococcal strain being reported with increasing frequency in recent years.3—6 However, the impact of drug-resistant S. pneumoniae, especially b—lactam resistance, is controversial as an international prospective study of 844 hospitalized patients failed to show a relationship between discordant therapy (i.e. inactive in vitro) and higher mortality rate.7

International surveillance studies play an important role in determining the extent of the resistance problem, charting the movement of important resistance phenotypes and genotypes across the globe, and monitoring the progress of intervention strategies designed to halt the progress of such strains.8,9 They can also help guide antibacterial selection and assist in the development of antibiotic usage policies and guidelines by providing regional information on the antibacterial susceptibilities of the key community-acquired RTI pathogens, and the likelihood of a resistant strain being encountered in a given patient. Although factors such as patient age, site of infection, and isolate source have been shown to influence the prevalence of resistance among S. pneumoniae reported in surveillance stu-dies,10—13 few large-scale studies have routinely collected and analyzed data according to demographic parameters. Against this background, PRO-TEKT (Prospective Resistant Organism Tracking and Epidemiology for the Ketolide Telithromycin), an international, longitudinal surveillance study, was initiated in 1999 to monitor the susceptibility of common respiratory bacterial pathogens from patients with community-acquired RTIs to current and new antibacterials, including the new ketolide telithromycin.

The study, which now includes 120 centers in over 30 countries, is routinely collecting demographic information, including patient age, gender, and in- and out-patient status, as well as isolate source and infection type. Particular attention is placed on the isolation of the three most important community-acquired RTI pathogens, namely Streptococcus pneumoniae, Haemophilus influenzae and Moraxella catarrhalis, along with Streptococcus pyogenes and Staphylococcus aureus. The antibac-

terial susceptibilities of these isolates according to geographic region have been presented in detail elsewhere.14-16

This paper reports the analysis of the impact of demographic variables on the antibacterial susceptibility, in particular to telithromycin, of S. pneumoniae isolates collected in the first year (1999-2000) of the PROTEKT study.

Methods

Organism collection, storage, and shipment

During the 1999—2000 respiratory season, 69 centers in 25 countries contributed isolates to the PROTEKT global study. These comprised 15 centers in North America, 13 in Latin America, 29 in Europe, nine in the Far East, and three in Australasia. Each center was requested to collect 60 isolates of S. pneumoniae from clinical samples obtained from patients with one of the following types of community-acquired RTIs: acute/chronic sinusitis, acute/ chronic otitis media, pharyngitis, community-acquired pneumonia (CAP), acute bacterial exacerbations of chronic bronchitis, or acute exacerbation of chronic obstructive airways disease.

The following culture sources were considered acceptable: blood, sputum (determined acceptable if the number of white blood cells exceeded the number of epithelial cells observed in the Gram stain), middle-ear fluid, bronchoalveolar lavage, nasopharyngeal swab or aspirate, and sinus aspirate. Only the first positive isolate from each patient was included in the study. Exclusion criteria included isolates from patients with cystic fibrosis, a nosocomial respiratory tract infection, or those who had been hospitalized for over 48 hours before the sample was taken, as well as duplicate strains and strains originating from existing collections. Patient demographic data, including patient age, gender, type of infection, culture source, in- or outpatient status, specimen accession number, collection date, but not previous antibiotic treatment, were requested for each isolate (data collected from laboratory records). Isolates were shipped to a central laboratory (GR Micro, London, UK) for analysis. Procedures for storage and shipment have been described in detail elsewhere.11

Microbiological and genetic analyses

To minimize the risk of phenotypic variation in the character of the stored isolate, isolates were

subcultured at the central laboratory, and then re-identified using standard procedures.11 Minimum inhibitory concentrations (MIC) were determined using the National Committee for Clinical Laboratory Standards (NCCLS) broth microdilution method17 and microplates containing a panel of lyophilised antibacterials (Sensititre, Trek Diagnostics Ltd. East Grinstead, UK).11 A multiplex rapid cycle polymerase chain reaction (PCR) method with microwell-format probe hybridization, as described previously,18 was used to test for the presence of erm(A), erm(A) subclass erm(TR), erm(B), erm(C), and mef(A) genes among macrolide-resistant S. pneumoniae.

Data analysis

MIC results were interpreted according to published NCCLS breakpoints (2002),19 where available. The recently approved NCCLS breakpoints (Fourteenth Informational Supplement M100-14, in press) of <1 mg/L susceptible, 2mg/L intermediate and >4 mg/L resistant were used for telithromycin.

Effects of country, culture source and patient characteristics (age, gender and in- or out-patient status) upon susceptibility data were analyzed using univariate comparison and multivariate linear regression. Patient age was divided into four category variables for analysis purposes: infants (children aged <2 years), children (aged 3-14 years), adults (aged 15-64 years) and elderly adults (aged >65 years). Infection type was not included in the analyses owing to the relatively low compliance to data collection (54.6% of patients). Univariate analyses were performed on penicillin, erythromycin, levofloxacin and telithromycin susceptibility results. Multivariate linear regression, to determine the odds of resistance to the given agent, was only performed on penicillin and erythromycin owing to the low levels of telithromycin and levofloxacin resistance. Categorical variables were used for country, patient age, culture source, gender, and treatment setting. Logistic regression (SAS v.8.02 [SAS Institute]) was used for multivariate analysis, Odds ratios (OR) and 95% confidence intervals (95% CIs) were calculated for those categories that showed a significant difference (P < 0.05) from the category baseline. The baseline categories were as follows: country, United Kingdom (UK); patient age, adults (15-64 years); culture source, blood; patient gender, male; treatment setting, in-patient.

The Hosmer-Lemeshow test of goodness of fit of the model was included. Individual pairwise comparisons were performed using either a Chi-square test or Fisher's exact test as appropriate.

Results

Origin of isolates

During the first year of PROTEKT, 3362 S. pneumoniae isolates from patients with community-acquired RTIs were tested. Information on patient age, culture source, patient gender, and treatment setting were available for 85.4%, 86.3%, 86.4%, and 85.6% of isolates, respectively. Of those isolates for which details of patient age were available, 15.3% were from infants, 12.6% from children, 42.2% from adults, and 29.8% from elderly adults. In adults and elderly adults, the primary culture source for S. pneumoniae was sputum (673/1198 [56.2%] and 635/846 [75.1%], respectively, among the isolates for which demographic data were available). In children, S. pneumoniae was isolated predominantly from sputum (79/354 [22.3%]) and the nasopharynx (73/354 [20.6%]), while in infants the ear (97/425 [22.8%]), nasopharynx (95/425 [22.4%]), and sputum (78/425 [18.4%]) were the predominant culture sources among the isolates for which demographic data were available.

Univariate analyses of the impact of demographic variables on in vitro activity and resistance rates

The antibacterial susceptibilities of isolates according to geographic region have been presented in detail elsewhere.14-16 The in vitro activities of penicillin, erythromycin, levofloxacin and telithromycin are shown as cumulative MIC distributions for a selection of culture sources in Figure 1a-d, and according to the patient age groups in Figure 2a-d. Telithromycin was the most potent of the antibacterials tested against S. pneu-moniae, irrespective of patient age or culture source, with the cumulative MIC distribution curve for telithromycin being markedly shifted to the left compared with the other agents. These cumulative MIC distributions indicate that there is a slight difference in the in vitro activity of teli-thromycin against isolates from the various culture sources and the various age groups, but this occurs only at low concentrations (<0.06 mg/L for patient age and <0.25 mg/L for culture source) after which points the curves converge and there appears to be no difference in telithromycin activity. In contrast, both penicillin and erythromycin show large separation in cumulative MIC distribution curves for the various culture sources and the various patient age groups across the whole concentration scale tested.

(a) Penicillin

cu 100

ri) 60

3 F 10

x^x—x—x—x—x

-ä-Ear

Nasopharyngeal

-x- Sputum

0.008 0 0150.03 0.06 0.12 0.25 0.5 1 2

MIC (mg/L)

16 32 64 128

JS o w

HI u (D

(b) Erythromycin

100 90 80 70 60 50 40 30 20 10 0

G :—x—J

-•- Blood

-ü-Ear

-♦- Nasopharyngeal

! -x- Sputum

0.03 0.06 0 IP 0.25 OS t £ 4 8 16 33

MIC (mg/L)

(c) Levofloxacin

<2 100 o

0 80 a

o> 70 n>

« 50 <u

—x—x—x

-t- Blood Ear

-♦- Nasopharyngeal -X- Sputum

OS 1 2 A

MIC (mg/L)

re is £ 3

(d) Telithromycin

7060 5040 3020 10-

o -l-x—>5-

O.CQ20.004 0.000 0.015 0.03 0.06 0.12 0 25 0.5 1 2

MIC (mg/L)

t/ s :—x—y—x—x—x R

-•- Blood

-6-Ear

-•- Nasopharyngeal

-X-Sputum

16 32 64 129

Figure 1 Cumulative MIC distribution curves of selected antibacterials against blood, ear, nasopharyngeal, and sputum isolates of Streptococcus pneumoniae (dotted lines represent NCCLS breakpoints). S, susceptible; I, intermediate; R, resistant (proposed NCCLS breakpoints for telithromycin).

MIC summary data and resistance rates for penicillin, erythromycin, levofloxacin, and telithromycin are summarized according to culture source (Table 1) and patient characteristics (Table 2). MIC summary data indicate that demographic variables appear to have little impact on in vitro activity of any of the above agents. In contrast, comparison of resistance rates suggests striking differences attributable to demographic variables. Penicillin resistance rates appeared highest among isolates from the ear (30.5%) and middle-ear fluid (26.7%; Table 1), while the middle-ear fluid and nasopharynx appeared to be major reservoirs of erythromycin (macrolide) resistance (43.3% and 40.6%, respectively). Pneumococcal isolates from sinus aspirates had the lowest rates of penicillin resistance (17.5%) of all the culture sources collected, yet high rates of erythromycin resistance (35.0%). Blood isolates were found to have low rates of resistance to both penicillin (18.7%) and erythromycin (21.3%). Fluor-oquinolone-resistant S. pneumoniae were isolated from the sinus (2.2%), throat (1.9%), sputum (1.7%), bronchoalveolar lavage samples (0.8%), and blood (0.4%). Only two isolates, one from bronchoalveolar

lavage and one from ear fluid, required telithromycin at >4 mg/L for inhibition.

Among S. pneumoniae isolates obtained from infants, 28.4% were penicillin-resistant, compared with 20.4% for isolates obtained from older children, 19.0% for isolates from adults, and 25.4% from elderly adults (Table 2). A similar resistance pattern was observed for erythromycin: 39.1% of pneumo-coccal isolates from infants were resistant to this agent, compared with 32.0%, 28.5% and 36.4% for isolates from older children, adults, and elderly adults, respectively. Most of the fluoroquinolone-resistant S. pneumoniae were recovered from elderly adults. There were no notable differences in antibacterial resistance patterns associated with the gender of the patients or with regard to in- or out-patient status.

For each culture source, there were marked differences in the prevalence of S. pneumoniae resistance to penicillin and to erythromycin among the age groups (Table 3). Among infants and adults, pneumococcal isolates from the ear, middle ear and sputum tended to have the highest prevalence of resistance to penicillin (infants: 39.2%, 37.5% and

<D D) (0 ■n C <1> o

'■J3 n 3

(a) Penicillin

100 90

70 60 50 -40 ■ 30 20 10 ■ 0

(b) Erythromycin

s 1 J y R

me/) -■- Adult

-o- Child

-•- Elderly

-A- Infant

O.0Ö8 0.015 0.03 0.06 0.12 0 2S 0.S I 2

MIC (mg/L)

3 c o o

s 1 R /

-o- Child

Elderly

■■■¿i Infant

16 32 64 128

0.CQ 0.06 0.12 0.25 0.5 1 2

MIC (mg/L)

8 16 32 64 128

(c) Levofloxacin

(d) Telithromyclri

v> <*—

o at a>

§ 50 | «0

s f 1 R

-m- Adult

-o- Child

Elderly

-6- Infant

0.5 I 2

MIC (mg/L)

16 32 64 123

to 100

ra 90 ■

o 70 ■

Gl CO 60

c Q 50

o 40 ■

£ 10 -

. s 1 R

-■-Adult

-o- Child

Elderly

-Cr- Infant

-Bt-O-t-1—,—,—,—.—,—.-

0.002 0 004 0.008 0.015 0.03 0 06 0LT2 0 25 0.5 1

S IS 32 f4 I2S

MIC (mg/L)

Figure 2 Cumulative MIC distribution curves of selected antibacterials against isolates of Streptococcus pneumoniae according to patient age group, infant 0—2 years, child 3—14 years, adult 15—64 years, elderly >65 years (dotted lines represent NCCLS breakpoints). S, susceptible; I, intermediate; R, resistant (proposed NCCLS breakpoints for telithro-mycin).

29.5%; adults: 24.3%%, 22.2% and 21.7%%, respectively), while in the elderly population, bronchoal-veolar lavage, sputum, and nasopharynx were the primary sources of penicillin-resistant S. pneumoniae (30.5%, 27.7% and 26.7%, respectively). Children had the highest prevalence of penicillin resistance among isolates from throat, nasopharynx and blood (26.9%, 24.7% and 24.4%, respectively). Rates of macrolide resistance exceeded 40% in S. pneumoniae cultured from middle-ear fluid, the nasopharynx, and the sinuses of infants; the nasopharynx and sinuses of older children; and isolates from bronchoalveolar lavage samples and sputum of elderly adults.

Multivariate analyses of the impact of demographic variables on resistance rates

Multivariate logistic regression models were fitted to determine which countries and demographic factors were primarily associated with the frequency of pneumococcal resistance to penicillin and erythro-

mycin. There were too few telithromycin-resistant and levofloxacin-resistant S. pneumoniae isolates to show any statistical difference in resistance rates between the country and demographic variables. Country, culture source and patient age were found to be significant predictors of resistance to penicillin, while country, patient age and gender were significant predictors of resistance to erythromycin (Table 4). Data on previous antibacterial therapy were not available for analyses.

The odds of penicillin resistance were 5.92-fold (Mexico) to 49.95-fold (South Korea) significantly higher in eight PROTEKT countries than in the UK (Table 4). Similarly, the odds of erythromycin resistance were 2.50-fold (Mexico) to 64.79-fold (South Korea) significantly higher in 11 PROTEKT countries than in the UK.

After adjustment for country and other demographic factors, ear swabs were the only culture source associated with significantly higher odds (OR 2.07) of penicillin resistance than blood isolates. None of the culture sources were associated with

u (C X

01 ■M

00 C-J a^

o o o o o cs o

00 vO cs cs n cs n 00 vO

01 Ofi m Lil m m m m m Lil

1 ooooooooo VI VI VI VI VI VI VI VI VI

-- -- -- -- -- --

n c* n o 00

cs CO vo n ri o Iii n Iii n c*

vO A vO A vO A vO A vO A ^T vO A ^T vO A vO A vO A

01 Ofi n o n O n O n O n C3 n C3 m C3 m m o

c (C R o VI o I o I o I c^ I c^ I c^ I o I c^ II

o u vO A vO A vO A vO A vO A vO A vO A vO A ^T vO A

O m u vO o vO o vO O vO O vO C3 vO C3 vO C3 vO o vO C3

o o o o c^ c^ c^

^ n Lil Lil a^

00 rj o n cs cs cs cs

A ^T A ^T A A A A

01 Ofi 00 o O 00 o O 00 o O m 00 o o 00 o C3 00 o o 00 o C3 00 o C3

1 o VI o I o I c^ c^ VI I c^ I I o VI

3E cs cs cs cs cs cs cs

o in u m O n O vO O m n o n o m o m m

o o o Lil c^ c^ c^ c^ o

o o o o o o o o m C-J o

■y 00 c^

01 Ofi cs o O o O cs o O o C3 00 o C3 00 o C3 o C3 00 o o o

Ü o VI o o VI c^ c^ c^ c^ c^ c^

o o U cs m CS m CS cs cs cs cs vO vO C3

3E o o o o o c^ c^ c^ o

o in u 00 o O m O m o 00 o C3 m m m C3 00 o m

2 o o o

cs 00 m o cs cs

o m o n n n 00 m vO

m cs cs cs

° < ¡5

CO CO LU

T3 01 IP

•S .5= a. -c

- U^ l/l '

effects upon the prevalence of erythromycin resistance.

After adjustment for other factors, pneumococcal isolates from infants had significantly higher odds of both penicillin resistance (1.98) and erythromycin resistance (1.89) than isolates from adults (Table 4); isolates from children and elderly adults were not significantly different to those from adults. Isolates from males had 0.69 times lower adjusted odds of erythromycin resistance than did isolates from females. Treatment setting was not associated with significant differences in the prevalences of penicillin or erythromycin resistance.

Impact of demographic variables on macrolide resistance mechanisms

Table 5 shows the distribution of macrolide resistance mechanisms among erythromycin-resistant S. pneumoniae, according to univariate analyses of demographic and clinical parameters. Erm(B) was the predominant mechanism of resistance across all age groups and, among sources with >10 erythro-mycin-resistant isolates, was particularly prevalent from bronchoalveolar lavage (69.1%) and sinus (68.8%) samples. Over 10% of S. pneumoniae isolated from infants or children were found to test positive for both erm(B) and mef (A) mechanisms; the majority of these isolates being collected in South Korea. Among the isolates from the ear or middle ear, the population carrying both erm(B) and mef (A) (17.3% or 15.4%, respectively) was high relative to the populations carrying either the erm(B) (42.7% or 53.8%, respectively) or mef (A) (37.3% or 30.8%, respectively) resistance mechanism alone. No differences in the distribution of macrolide resistance mechanisms were associated with gender or inpatient/outpatient setting.

Multivariate analyses were not undertaken due to the complexity of the model required to accommodate at least four genotype groups.

Discussion

In response to the worldwide problem of antibacterial resistance, strategic programmes for controlling the emergence and spread of antibacterial resistance are being implemented at international, national, and regional levels.20-22 Resistance surveillance studies are an integral component of many of these strategies.9,23 For surveillance studies to be of clinical value to the physician making choices about empirical antibacterial therapy, ideally they should collect isolates from specific infection sites in carefully defined patient populations.9 Collection of

Table 2 In vitro activity of selected antibacterials against Streptococcus pneumoniae (n = 3362): analysis by patient age group, gender, and treatment setting.

No. of isolates Telithromycin Penicillin Erythromycin Levofloxacin

MIC (mg/L) %R1 MIC (mg/L) %R2 MIC (mg/L) %R2 MIC (mg/L) %R2

MIC50 MIC90 Range MIC50 MIC90 Range MIC50 MIC90 Range MIC50 MIC90 Range

Age group (years)

0—2 440 0.015 0.25 <0.002- 1 0 0.06 4 <0.008- >4 28.4 0.06 >64 <0.03- >64 39.1 1 1 <0.5- 2 0.0

3—14 362 0.015 0.12 <0.002- 0.5 0 0.03 2 <0.008- 4 20.4 0.06 >64 <0.03- >64 32.0 1 1 <0.5- 8 0.3

15—64 1213 0.015 0.12 0.004- 8 0.1 0.015 2 <0.008- >4 19.0 0.06 >64 <0.03- >64 28.5 1 1 <0.5- 32 1.0

>65 857 0.015 0.12 0.004- 4 0.1 0.03 2 <0.008- >4 25.4 0.06 >64 <0.03- >64 36.4 1 1 <0.5- 32 2.3

Unspecified 490 0.015 0.06 0.004- 0.5 0 0.03 2 <0.008- >4 19.6 0.006 >64 <0.03- >64 19.8 1 1 <0.5- 16 0.4

Gender

Male 1831 0.015 0.12 0.004- 8 0.1 0.03 2 <0.008- >4 24.2 0.06 >64 <0.03- >64 32.0 1 1 <0.5- 32 1.2

Female 1073 0.015 0.25 <0.002- 1 0 0.015 2 <0.008- >4 21.0 0.06 >64 <0.03- >64 32.8 1 1 <0.5- 16 0.9

Unspecified 458 0.015 0.12 0.004- 0.5 0 0.03 2 <0.008- >4 16.6 0.06 >64 <0.03- >64 22.9 1 1 <0.5- 16 0.7

Treatment setting

Inpatient 1452 0.015 0.12 0.004- 4 0.1 0.015 2 <0.008- >4 23.1 0.06 >64 <0.03- >64 31.7 1 1 <0.5- 32 1.0

Outpatient 1425 0.015 0.12 <0.002- 8 0.1 0.03 2 <0.008- >4 23.9 0.06 >64 <0.03- >64 35.1 1 1 <0.5- 32 1.2

Unspecified 483 0.015 0.06 0.004- 0.5 0 0.03 2 <0.008- >4 13.9 0.06 16 <0.03- >64 17.0 1 1 <0.5- 16 0.6

R1 = resistant according to proposed NCCLS breakpoint (telithromycin-resistant, MIC >4 mg/L); R2 = resistant according to NCCLS breakpoints (penicillin-resistant, MIC >2 mg/L; erythromycin-resistant, MIC >1 mg/L; levofloxacin-resistant, MIC >8 mg/L).

o cr 0)

e t la ol

e t la ol

.9) 3) .4)

7. .7) .6) 0) .0) 0) 0. 0) .2) 6.

0. 0. (4 8. 3

( (4 8. (4 1 (

n (2 (0. (0. 5 (0. ( 7

2 9 5 3 5

1 5 7 2 1 1 6 4 1 8

2/ 4/ 2/ 0/ 6/ 0/ 6/ 0/ 2/ 2/

2 2 5 1

) 3) ) ) 3)

.5) ) ) ) ) )

3. .8 .5 .0 .7 .7

3 0.

cs 3 7. 7. 0. 3 6. 6.

—' ( 3 3. 2 2 ( 1 2

7 4 ( 3 ( ( 3 ( (

2 2 7 1 5 7 2 5

2 1 3 9 5 6 6 1 1

3/ 9/ 4/ 3/ 4/ 3/ 4/ 2/ 4/

5 3 1 0

.6) .2) .6) .8) .2) .0) .5)

14. 4. 1. .0) 3. 5. 8. 1.

2 3 (4 (4 3 1

1 3 7 3 2 9 6

4 3 5 3 7 4 7 2

6/ 8/ 8/ 0/ 2/ 3 9/ 0/ 3 3/

807 729

788 13

0) 13.0 0.5)

3 ( (1 (0 ( 2 9 ( ( 5

in N Nt-

.7 7. )

O S o do

0. 5 (0. 3(

.7 0) 6.

o O vO o

.5 .1 4. 6.

.4) .2) 3)

4. (2 5. 9.

41 3 3 7 5

0/ 5/ 11/

-—(VI —• -— 7 ( 15

3956 9/ 2/ 8/ 0/

.7) .8) .5) .9)

4. (2 3. (2 6. 6. (2

3 7 2 4 9 7 6 2

8/ 0/ 3/ 7/

.2) .0) .2) .5) .4) .8) .5) .0) 3)

c* 0. (2 9. 3 7. 3 8. (2 3. 9. (2 5. (2 3. 3

8 7 0 2 7 9 6 5 9 9 2 8 7 2 5

5/ 4/ 8/ 3 6/ 7/ 2 4/ 3/ 2 3/ 5/

r yn a ry

.5) 3) .2) .2) 3) .7) 9. 3 0

2. (6 5. (4 5. (5 7. 3 3. 3 6. (4

6 5 9 9 2 8 7 2 5 4 /4

0/ 3/ 4 6/ 9/ 2 4/ 7/ 2/ 7

) .0)

.7 ) 9.

2 3. (

( 8. ( 1 ( 3

3 21

6 1 /1 1

6/ 1/ 2/ 1/

14 3

d e fii

h m t ci d le p u at e od dl op us tu oa p

lo AL ar id as in pu hr ns maiLjSZi/iwl-3

demographic and source data (such as patient age, culture source, infection type, community- or hospital-acquired infection), which are often missing from surveillance reports, and the ability to analyze the data according to these parameters are important to reduce bias and error arising from variation in the sample population, and to answer the question of relevance to the physician.8 Moreover, given the unpredictability of resistance development, the more information available on the potential population sources of significant new resistance developments, the more likely are the chances to introduce appropriate interventions in a timely fashion to limit the spread of the resistant pathogens and reduce any potential adverse health impact.

A number of international surveillance studies have tracked the susceptibility of RTI pathogens (e.g., Alexander, SENTRY).1,24 While these studies provide a wealth of useful information, one of their major drawbacks has been in the limited collection of demographic and clinical data.8 This need has been addressed in the design of the PROTEKTstudy, which collects information on patient age, gender, type of infection, culture source, and in- or outpatient status from patients with community-acquired RTIs. In the first year report from the study on S. pneumoniae, compliance with demographic data on culture source, patient age, patient gender, and treatment setting was high (>85%). Compliance with the collection of demographic data and the integrity of the data collected was encouraged by the use of electronic spreadsheets with preset choices for selection, by limiting the amount of data that investigators needed to collect to fulfill the study objectives, and by the use of a detailed study protocol. To the authors' knowledge this is the most comprehensive collection of demographic and clinical data as part of a large-scale international resistance surveillance study of respiratory pathogens to date.

To ensure patient confidentiality, demographic and clinical information were not collected from source, but 'second hand' from laboratory records. This has obvious limitations in terms of the type of data that can be collected, as well as in the completeness of the data. In this study, completion of information on disease type was low compared with the other demographic parameters, and this parameter was therefore not included as a variable in the logistic regression analysis. The low compliance with the collection of data on disease type might reflect in part problems of 'second-hand' data collection, as well as concerns regarding identification of individual patients via collecting certain types of information. A more complete picture of the influence of demographic and clinical para-

Table 4 Results of multivariate logistic regression analysis on prevalence of penicillin resistance and erythromycin

resistance among Streptococcus pneumoniae (n = 3362) isolated from patients with community-acquired respiratory

tract infections during 1999-2000.

Penicillin resistancea Erythromycin resistance b

Odds ratioc (95% CI) P value Odds ratioc (95% CI) P value

Country

UKd 1 (-) — 1 (-) —

Belgium 3.13 (1.06-9.22) 0.0384

Eire 11.09 (3.75-32.78) <0.0001 3.20 (1.33-7.73) 0.0096

France 17.90 (6.80-47.09) <0.0001 10.84 (5.39-21.78) <0.0001

Hong Kong 27.85 (9.88-78.50) <0.0001 22.07 (9.72-50.07) <0.0001

Hungary - NS 9.99 (4.36-22.09) <0.0001

Italy - NS 5.78 (2.80-11.92) <0.0001

Japan 15.50 (6.05-39.77) <0.0001 28.85 (14.52-57.36) <0.0001

Mexico 5.92 (2.15-16.35) 0.0006 2.50 (1.18-3.87) 0.0168

S. Korea 49.95 (18.56-134.50) <0.0001 64.79 (28.62-146.67) <0.0001

Spain 9.82 (3.44-28.05) <0.0001 4.46 (2.00-9.94) 0.0003

USA 10.41 (4.00-27.12) <0.0001 3.41 (1.72-6.75) 0.0004

Culture source

Blood d 1 (-) — 1 (-) —

Ear 2.07 (1.27-3.38) 0.0037 - NS

Age (years)

0-2 d 1.98 (1.42-2.77) <0.0001 1.89 (1.37-2.60) <0.0001

3-14 1.09 (0.74-1.61) NS 0.98 (0.69-1.41) NS

15-64 1 (-) — 1 (-) —

>65 1.14 (0.88-1.47) NS 1.01 (0.79-1.28) NS

Gender

Maled 1 (-) — 1 (-) —

Female - NS 0.69 (0.56-0.84) 0.0003

Treatment setting

Inpatientd 1 (-) — 1 (-) —

Outpatient - NS - NS

BAL = bronchoalveolar lavage; CI = confidence intervals.

a Penicillin-resistant, MIC >2 mg/L.

b Erythromycin-resistant, MIC >1 mg/L.

c Odds ratios calculated only for those groups found to be significantly different from the reference group.

d Reference group.

meters on resistance patterns could be obtained by collecting data on these parameters at source. However, to run resistance surveillance as a true epide-miological/clinical study would have significant confidentiality and cost implications and necessitate collection of informed patient consent, making it impractical for large-scale international surveys, such as the present study.

Despite the limitations of 'second-hand' data collection, this study has highlighted some striking and highly significant relationships between resistance prevalence in S. pneumoniae and demographic parameters using both univariate and multivariate analyses. Univariate analyses indicated marked effects of patient age and culture source; however such analyses do not take into account local or national prescribing practices, resistance levels and culture sampling procedures (e.g. tym-

panocentesis) which may introduce a bias into the results. Using a multivariate logistic regression model, it was found that country, culture source (ear swabs) and patient age were significant predictors of resistance to penicillin, while country, patient age and gender were significant predictors of resistance to erythromycin. The prevalence of penicillin and/or erythromycin resistance was significantly higher in 18 of the participating countries other than in Germany. The multivariate logistic regression model also enables quantifying the magnitude of the effect of each variable using an odds ratio. The odds ratios associated with these countries were higher than any of the odds ratios determined for culture source, patient age and patient gender. Among culture sources, only ear swabs, with an odds ratio of 2.07, were significantly more likely to exhibit penicillin resistance than blood cultures.

Table 5 Distribution of macrolide resistance mechanisms among erythromycin-resistant Streptococcus pneumoniae

according to demographic parameters.

ERY-R Genotype

n erm(B) erm(A) subclass mef(A) mef(A) & erm(B) Negative

erm (TR)

n (%) n (%) n (%) n (%) n (%)

Culture source

Blood 107 55 (51.4) 0 (0.0) 44 (41.1) 5 (4.7) 3 (2.8)

BAL 81 56 (69.1) 0 (0.0) 21 (25.9) 3 (3.7) 1 (1.2)

Ear 75 32 (42.7) 1 (1.3) 28 (37.3) 13 (17.3) 1 (1.3)

Middle-ear fluid 13 7 (53.8) 0 (0.0) 4 (30.8) 2 (15.4) 0 (0.0)

Nasopharynx 95 49 (51.6) 0 (0.0) 37 (38.9) 8 (8.4) 1 (1.1)

Sinus 48 33 (68.8) 0 (0.0) 13 (27.1) 2 (4.2) 0 (0.0)

Sputum 526 304 (57.8) 0 (0.0) 178 (33.8) 33 (6.3) 11 (2.1)

Throat 9 6 (66.7) 0 (0.0) 2 (22.2) 1 (11.1) 0 (0.0)

Age (years)

0-2 172 82 (47.7) 0 (0.0) 71 (41.3) 18 (10.5) 1 (0.6)

3-14 116 63 (54.3) 1 (0.9) 39 (33.6) 12 (10.3) 1 (0.9)

15-64 346 202 (58.4) 0 (0.0) 111 (32.1) 22 (6.3) 11 (3.2)

>65 312 183 (58.7) 0 (0.0) 107 (34.3) 17 (5.4) 5 (1.6)

Gender

Male 586 345 (58.9) 0 (0.0) 170 (29.0) 44 (7.5) 7(1.2)

Female 352 184 (52.3) 1 (0.3) 135 (38.4) 25 (7.1) 7 (2.0)

Treatment setting

Inpatient 461 276 (59.9) 0 (0.0) 149 (32.3) 28 (6.1) 8 (1.7)

Outpatient 500 268 (53.6) 1 (0.2) 180 (36.0) 41 (8.2) 10 (2.0)

BAL = bronchoalveolar lavage; ERY-R = erythromycin-resistant (MIC >1 mg/L).

It was also found that the adjusted odds of penicillin resistance and erythromycin resistance among isolates from infants to be significantly higher (1.98 and 1.89, respectively) than among isolates from adults. These findings are consistent with those from other studies reported in the literature.10,12,25-27 Among children aged <6 years, Arason and colleagues found by multivariate analysis that age <2 years significantly increased the odds of carrying penicillin-resistant pneumococci (by 3.5-4-fold compared with children aged 2-6 years).25 Similarly, Weiss et al. have recently shown in a surveillance study of pneumococci circulating in Quebec during the 2000-2001 winter season that resistance to macrolides and to penicillin was higher among invasive than non-invasive isolates, and among pediatric isolates compared with those obtained from adults.27 In their study, Weiss et al. failed to find any significant relationship between pneumo-coccal resistance and patient gender.27 In the present study, however, a lower odds ratio (0.69) of pneumococcal erythromycin resistance in males than in females was found.

Erm(B) was the predominant mechanism of macrolide resistance among S. pneumoniae across all age groups and culture sources, and was particularly prevalent among S. pneumoniae isolates

obtained from bronchoalveolar lavage and sinus samples. The product of this gene modifies the ribosomal target site of the macrolides by dimethy-lating a specific adenine residue on the 23S rRNA (A2058 - Escherichia coli numbering). The resulting conformational change in the ribosome produces decreased binding of all macrolide, lincosamide and streptogramin B antibacterials (the so-called MLSB phenotype), and high-level resistance to these

agents.

28—30

Erm(A) subclass erm(TR) has also

recently been reported in S. pneumoniae,31 and was found in one macrolide-resistant pneumococcal isolate in the present study, from an ear sample obtained from a three-year-old girl in Australia. Overall, 35.0% (365/1043) of macrolide-resistant S. pneumoniae in the present study tested positive for the product of the mef (A) gene, indicating that macrolide resistance in these isolates was due to drug efflux,29 and 6.8% (71/1043) tested positive for both erm(B) and mef(A). This dual resistance mechanism, which generally causes high-level MLSB resistance, was particularly prevalent among children and among isolates from the ear/middle-ear fluid [20.8% (10/48) of macrolide-resistant S. pneumoniae isolated from ear/middle-ear fluid samples from children <2 years tested positive for this dual mechanism of resistance in the present study].

Further investigation of 16 of the isolates that tested negative for the mechanisms studied, found all were susceptible to telithromycin and that macrolide resistance was conferred by ribosomal

gene mutations.32

The fluoroquinolones showed different patterns of pneumococcal resistance to the other agents, reflecting differences in their licensed indications including the absence of pediatric usage. Levoflox-acin resistance was extremely rare among S. pneu-moniae isolated from children, as anticipated, and no levofloxacin-resistant pneumococci were isolated from the ear/middle-ear fluid or from the nasopharynx. The highest rates of levofloxacin-resistant pneumococci were reported in elderly adults (2.3%) and in isolates obtained from the sinus (2.2%), throat (1.9%) and from sputum (1.7%), although the small numbers of levofloxacin-resis-tant isolates collected precluded statistical analysis.

Antibiotic consumption and its impact upon antimicrobial resistance is not included in the PROTEKT study. However, studies conducted elsewhere have shown a strong correlation between high antibiotic consumption and the highest rates of selection of resistant organisms at patient, community, region or country level, prompting recommendations for greater controls on antibiotic usage.34-36

One of the objectives of the PROTEKT study was to provide current surveillance data for the ketolide telithromycin at its time of launch into clinical practice and to track its susceptibility with time. Previously reported data from PROTEKT have shown that this ketolide has good in vitro activity against bacterial pathogens implicated in community-acquired RTIs, including S. pneumoniae isolates that are resistant to penicillin, macrolides, and/or fluoroquinolones.33 Only two (0.06%) isolates were resistant to telithromycin (MIC 4 and 8 mg/L). Both were multi-resistant isolates, carrying the erm(B) mechanism of macrolide resistance, and associated with different demographic variables (one isolate from bronchoalveolar lavage fluid, elderly male, France; the other from ear culture, adult male, Italy). Overall, this present analysis found that the high in vitro activity of telithromycin was largely unrelated to any of the demographic variables collected.

In summary, this analysis of the PROTEKT 19992000 database has shown some striking and highly significant differences in resistance prevalence among the key respiratory bacterial pathogens when analyzed according to country, demographic and clinical parameters, and highlights the importance of collecting patient data routinely as part of antibacterial surveillance.

Acknowledgements

We wish to thank the PROTEKT investigators around the globe who provided clinical isolates for this study. We also acknowledge the researchers at GR Micro Ltd (London, UK) for their analysis of the isolates, and Micron Research (Cambridgeshire, UK) for data analysis. The PROTEKT study is supported by Aventis.

Conflict of interest: No conflict of interest to declare.

References

1. Felmingham D, Washington J. Trends in the antimicrobial susceptibility of bacterial respiratory tract pathogens-findings of the Alexander Project 1992—1996. J Chemother 1999;11:5-21.

2. Felmingham D, GrUneberg RN, the Alexander Project Group. The Alexander Project 1996—1997: latest susceptibility data from this international study of bacterial pathogens from community-acquired lower respiratory tract infections. J Antimicrob Chemother 2000;45:191-203.

3. Dagan R, Klugman KP, Craig WA, Baquero F. Evidence to support the rationale that bacterial eradication in respiratory tract infection is an important aim of antimicrobial therapy. J Antimicrob Chemother 2001;47:129-40.

4. Fogarty C, Goldschmidt R, Bush K. Bacteremic pneumonia due to multidrug-resistant pneumococci in 3 patients treated unsuccessfully with azithromycin and successfully with levo-floxacin. Clin Infect Dis 2000;31:613-5.

5. Davidson R, Cavalcanti R, Brunton JL, Bast DJ, de Azavedo JCS, Kibsey P, Fleming C, Low DE. Resistance to levofloxacin and failure of treatment of pneumococcal pneumonia. N Engl J Med 2002;346:747-50.

6. Lonks JR, Garau J, Gomez L, Xercavins M, de Echanguen O, Gareen IF, Reiss PT, Medeiros AA. Failure of macrolide antibiotic treatment in patients with bacteremia due to erythro-mycin-resistant Streptococcus pneumoniae. Clin Infect Dis 2002;35:556-64.

7. Yu VL, Chiou CC, Feldman C, Ortqvist A, Rello J, Morris AJ, et al. International Pneumococcal Study Group. An international prospective study of pneumococcal bacteremia: correlation with in vitro resistance, antibiotics administered, and clinical outcome. Clin Infect Dis 2003;37:230-3.

8. Felmingham D. The need for antimicrobial resistance surveillance. J Antimicrob Chemother 2002;50:1-7.

9. Masterton RG. Surveillance studies: how can they help in the management of infection? J Antimicrob Chemother 2000;46: 53-8.

10. Sahm DF, Jones ME, Hickey ML, Diakun DR, Mani SV, Thorns-berry C. Resistance surveillance of Streptococcus pneumo-niae, Haemophilus influenzae and Moraxella catarrhalis isolated in Asia and Europe, 1997-1998. J Antimicrob Chemother 2000;45:457-66.

11. Gur D, Guciz B, Hascelik G, Esel D, Sumerkan B, Over U, Soyletir G, Ongen B, Kaygusuz A, Toreci K. Streptococcus pneumoniae resistance in Turkey. J Chemother 2001; 13: 541 -5.

12. Albanese BA, Roche JC, Pass M, Whitney CG, McEllistrem MC, Harrison LH. Geographic, demographic, and seasonal differ-

ences in penicillin resistant Streptococcus pneumoniae in Baltimore. Clin Infect Dis 2002;34:15-21.

13. Laurichesse H, Romaszko JP, Nguyen LT, Souweine B, Poirier V, Guelon D, Andre M, Ruivard M, De Champs C, Caillaud D, Labbe A, Beytout J. Clinical characteristics and outcome of patients with invasive pneumococcal disease, Puy-de-Dome, France, 1994-1998. Eur J Clin Infect Dis 2001;20:299-308.

14. Canton R, Loza E, Morosini MI, Baquero F. Antimicrobial resistance amongst isolates of Streptococcus pyogenes and Staphylococcus aureus in the PROTEKTantimicrobial surveillance programme during 1999-2000. J Antimicrob Chemother 2002;50:9-24.

15. Felmingham D, Reinert RR, Hirakata Y, Rodloff A. Increasing prevalence of antimicrobial resistance among isolates of Streptococcus pneumoniae from the PROTEKT surveillance study, and comparative in vitro activity of the ketolide, telithromycin. J Antimicrob Chemother 2002;50:25-37.

16. Hoban D, Felmingham D. The PROTEKT surveillance study: antimicrobial susceptibility of Haemophilus influenzae and Moraxella catarrhalis from community-acquired respiratory tract infections. J Antimicrob Chemother 2002;50:49-59.

17. National Committee for Clinical Laboratory Standards. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; approved standard, fifth edition. NCCLS document M7-A5 (ISBN 1-56238-394-9). NCCLS 2000, Wayne, PA 19087-1898, USA.

18. Farrell DJ, Morrissey I, Bakker S, Felmingham D. Detection of macrolide resistance mechanisms in Streptococcus pneumoniae and Streptococcus pyogenes using a multiplex rapid cycle PCR with microwell-format probe hybridisation. J Antimicrob Chemother 2001;48:541-4.

19. National Committee for Clinical Laboratory Standards. Performance standards for antimicrobial susceptibility testing; twelfth informational supplement. NCCLS document M100-12 (ISBN 1-56238-454-6) NCCLS2002, Wayne, PA 19087-1898, USA.

20. European Commission. Communication from the commission on a community strategy against antimicrobial resistance [COM (2001) 333]. Brussels; Commission of the European Communities. (http://europa.eu.int).

21. World Health Organization. Containing antimicrobial resistance: review of the literature and report of a WHO workshop on the development of a global strategy for the containment of antimicrobial resistance [WHO/CDS/CSR/DRS99.2]. Geneva, 1999. (www.who.int/emc).

22. Observatoire Nationale de l'Epidemiologie de la Résistance Bactérienne aux Antibiotiques. Recommendations methodologiques pour la surveillance de la résistance aux antibiotiques dans les laboratoires de microbiologie. 2002. (www.onerba.org).

23. Marchese A, Schito GC. Role of global surveillance in combating bacterial resistance. Drugs 2001;61:167-73.

24. Hoban DJ, Doern GV, Fluit AC, Roussel-DelvallezM, Jones RN. Worldwide prevalence of antimicrobial resistance in Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis in the SENTRY antimicrobial surveillance programme, 1997-1999. Clin Infect Dis 2001;32:S81-93.

25. Arason VA, Kristinsson KG, Sigurdsson JA, Stefansdottir G, Molstad S, Gudmundsson S. Do antimicrobials increase the carriage rate of penicillin resistant pneumococci in children? Cross sectional prevalence study BMJ 1996;313:387-91.

26. Doern GV, Brueggemann A, Holley Jr AP, Rauch AM. Antimicrobial resistance of Streptococcus pneumoniae recovered from outpatients in the United States during the winter months of 1994 to 1995: results of a 30-center national surveillance study. Antimicrob Agents Chemother 1996;40: 1208-13.

27. Weiss K, Guilbault C, Cortes L, Restieri C, Low DE. Genotypic characterization of macrolide-resistant strains of Streptococcus pneumoniae isolated in Quebec, Canada, and in vitro activity of ABT-773 and telithromycin. J Antimicrob Chemother 2002;50:403-6.

28. Fernandez-Munoz R, Monro RE, Torres-Pinendo R, Vasquez D. Substrate- and antibiotic-binding sites at the peptidyl-trans-ferase center of Escherichia coli ribosomes. Studies on the chloramphenicol, lincomycin and erythromycin sites. Eur J Biochem 1971;23:185-93.

29. Roberts MC, Sutcliffe J, Courvalin P, Bogo Jensen L, Rood J, Seppala H. Nomenclature for macrolide and macrolide-lin-cosamide-streptogramin B resistance determinants. Antimicrob Agents Chemother 1999;43:2823-30.

30. Weisblum B. Erythromycin resistance by ribosome modification. Antimicrob Agents Chemother 1995;39:77-585.

31. Syrogiannopoulos GA, Grivea IN, Tait-Kamradt A, Katopodis GD, Beratis NG, Sutcliffe J, Appelbaum PC, Davies TA. Identification of an erm(A) erythromycin resistance methylase gene in Streptococcus pneumoniae isolated in Greece. Antimicrob Agents Chemother 2001;45:342-4.

32. Farrell DJ, Douthwaite S, Morrissey I, Bakker S, Poehlsgaard J, Jakobsen L, Felmingham D. Macrolide resistance by ribo-somal mutation in clinical isolates of Streptococcus pneumoniae from the PROTEKT 1999/2000 study. Antimicrob Agents Chemother 2003;47:1777-83.

33. Felmingham D, Zhanel G, Hoban D. Activity of the ketolide telithromycin against typical community-acquired respiratory pathogens. J Antimicrob Chemother 2001;48:33-42.

34. Cars O, Mölstrad S, Melander A. Variation in antibiotic use in the European Union. Lancet 2001;357:1851-3.

35. Baquero F, Baquero-Artigao G, Canton R, Garcia-Rey C. Antibiotic consumption and resistance selection in Streptococcus pneumoniae. J Antimicrob Chemother 2002;50:27-37.

36. Cizman M. The use and resistance to antibiotics in the community. Int J Antimicrob Agents 2003;21:297-307.

Available online at www.sciencedirect.com

SCIENCE ^DIRECT«