A

hsso1fi©

K

1

auomyNiad

0d

International Journal of Antimicrobial Agents 28 (2006) 166–171

Decreasing prevalence of antimicrobial resistance in non-typhoidalSalmonella isolated from children with bacteraemia

in a rural district hospital, Kenya

Samuel Kariuki a,b,∗, Gunturu Revathi c, John Kiiru a, Brett Lowe d,e,James A. Berkley d,e, C. Anthony Hart b

a Centre for Microbiology Research, Kenyatta National Hospital, P.O. Box 20723, Nairobi, Kenyab Department of Medical Microbiology and Genito-Urinary Medicine, University of Liverpool, Liverpool L69 3GA, UK

c Department of Medical Microbiology, Kenyatta National Hospital, P.O. Box 20723, Nairobi, Kenyad Centre for Geographic Medicine, Kenya Medical Research Institute, P.O. Box 231, Kilifi, Kenya

e Centre for Clinical Vaccinology and Tropical Medicine, University of Oxford, Old Road, Oxford OX3 7LJ, UK

Received 16 February 2006; accepted 14 May 2006

bstract

We analysed 336 non-typhoidal Salmonella (NTS) isolated from children <13 years of age with bacteraemia admitted to a rural districtospital in Kenya from 1994 to 2005. Pulsed-field gel electrophoresis was used to determine genetic relatedness of strains, and antimicrobialusceptibility testing was also performed. Most NTS were either Salmonella enterica serovar Typhimurium (n = 114; 33.9%) or S. entericaerovar Enteritidis (n = 128; 38.1%), with minimal genotypic diversity over the study period. The NTS showed a remarkable decrease in levels

f resistance especially to two commonly available antimicrobials (amoxicillin and co-trimoxazole), from highs of 69.2% and 68.4% during994–1997 to 11% and 13%, respectively, in 2002–2005 (P < 0.01). All NTS remained fully susceptible to cefotaxime and ciprofloxacin. Ourndings show that commonly available drugs may still be useful for treatment of invasive NTS infections in this rural population.2006 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.

esistanc

ip

oiirdft

eywords: Non-typhoidal Salmonella; Bacteraemia; Children; Antibiotic r

. Introduction

Although non-typhoidal Salmonella (NTS) infections arecommon cause of self-limiting diarrhoea in healthy individ-als, NTS bacteraemia with or without focal infections canccur as a complication [1–3]. In Africa, NTS is among theost common causes of bloodstream infections in children

ounger than 5 years of age [1,4–6]. For instance, in Kenya,TS are second only to invasive pneumococcal disease in

mportance as a cause of bacteraemia in children <5 years of

ge, with a high mortality, especially in malnourished chil-ren. We have previously estimated that the local minimum

∗ Corresponding author. Tel.: +254 202 718 247; fax: +254 202 711 673.E-mail address: skariuki@kemri.org (S. Kariuki).

Ioqap2

924-8579/$ – see front matter © 2006 Elsevier B.V. and the International Societyoi:10.1016/j.ijantimicag.2006.05.026

e; Kenya

ncidence of community-acquired NTS is 166 per 100 000er year for children <5 years of age [6,7].

Early identification and effective antimicrobial treatmentf cases is an important step in the management of youngnfants with invasive bacterial infections. Currently, theres an increasing number of reports of cases of multidrug-esistant food-borne Salmonella enterica infections both ineveloped and developing countries, with few options leftor antimicrobial treatment. Examples of increases in resis-ance in NTS in developing countries, particularly in thendian subcontinent and Southeast Asia, are exemplified byutbreaks caused by multidrug-resistant NTS. These are fre-

uently resistant to the newer quinolones both in hospitalsnd the community over wide geographical areas [8,9]. Inrevious studies in Kenya over a 10-year period from 1994 to003, we also observed a steady increase in the proportion of

of Chemotherapy. All rights reserved.

l of Ant

mftrct3iwtcs

2

2

ibssosstctotbppsI9racan

acattiS

2p

a

pnpt1td1svrta2an

2

c((cc(omtdactavS

2

tvU

3

3

es

S. Kariuki et al. / International Journa

ultidrug resistance among NTS isolated from adult patientsrom a major urban population admitted to hospital with bac-eraemia [10]. In that study, the prevalence of NTS multiplyesistant to all commonly available drugs including ampi-illin, streptomycin, sulphamethoxazole/trimethoprim (co-rimoxazole), chloramphenicol and tetracycline rose from1% in 1994 to 42% in 2003, with concomitantly higher min-mum inhibitory concentrations (MICs) for each drug. Heree report on the observation of decreasing levels of resistance

o commonly used antimicrobials among NTS isolated fromhildren with bacteraemia admitted to a local rural hospitalituated along the coast of Kenya between 1994 and 2005.

. Materials and methods

.1. Study patients and bacterial isolates

The study site was Kilifi District Hospital (KDH), whichs a rural, government-funded hospital with 42 paediatriceds and an average of 5000 paediatric admissions per yearerving a local population of ca. 240 000, mainly small-cale farmers and fishermen. Before August 1998, culturesf cerebral spinal fluid and blood were initiated on clinicaluspicion of meningitis or sepsis; thereafter, all acute admis-ions, except those admitted for observation following minorrauma and elective admissions, were investigated with bloodultures. Written consent for participation of the children inhe study was given either by parents or a guardian. Treatmentf inpatients at KDH is based on World Health Organiza-ion guidelines [10]. Infants <60 days old are treated withenzylpenicillin plus gentamicin. Older children with sus-ected meningitis or septic shock are given benzylpenicillinlus chloramphenicol. Amoxicillin or benzylpenicillin is pre-cribed alone to treat mild or severe pneumonia, respectively.n children with prostration or pulse oximetry readings below0% or not able to drink (very severe pneumonia), chlo-amphenicol is substituted. For children with malnutrition,mpicillin and gentamicin are used. Co-trimoxazole is theommonest outpatient antibiotic in Kilifi district and is alsovailable in some chemist shops without prescription, but isot used for inpatients.

A total of 336 invasive NTS isolates were obtained overperiod of 12 years (1994–2005) from blood cultures of

hildren aged 0–13 years (75% of these were <5 years ofge). NTS isolates were identified by serotyping followinghe Kauffmann–White scheme utilising Salmonella agglu-ination antisera (Murex Diagnostics, Dartford, UK). NTSsolates were stored at −80 ◦C on Protect beads (Technicalervice Consultants Ltd., Heywood, UK) until analysed.

.2. Characterisation of NTS by pulsed-field gel electro-

horesis (PFGE) of macrorestricted chromosomal DNA

Chromosomal DNA from NTS isolates was prepared ingarose plugs as described previously [11]. DNA in agarose

tso1

imicrobial Agents 28 (2006) 166–171 167

lugs was digested using 25 U of XbaI or SpeI (Roche Diag-ostics GmbH, Mannheim, Germany). PFGE of agaroselug inserts was then performed on a CHEF-DR III sys-em (Bio-Rad Laboratories, Hercules, CA) on a horizontal% agarose gel for 20 h at 120 V, with a pulse time of 1 so 40 s at 14 ◦C. A lambda DNA digest consisting of a lad-er (ca. 22 fragments) of increasing size from 50 kb to ca.000 kb was included as a DNA size standard. The gel wastained with ethidium bromide and photographed on an ultra-iolet transilluminator (UVP Inc., San Gabriel, CA). Theestriction endonuclease digest patterns were compared andheir similarities were scored by the method of Tenover etl. [12] and using the Dice similarity coefficient formula,h/(a + b), where h is the number of matching bands and+ b is the total number of bands including matching andon-matching.

.3. Antibiotic susceptibility testing

NTS were tested for susceptibility to antimicrobials by aontrolled disk diffusion technique on Mueller–Hinton agarOxoid Ltd., Basingstoke, UK) plates. The antibiotic disksall from Oxoid Ltd.) contained amoxicillin (10 �g), tetra-ycline (30 �g), co-trimoxazole (1/25 �g), chlorampheni-ol (30 �g), gentamicin (10 �g), amoxicillin/clavulanic acidco-amoxiclav) (20/10 �g), ciprofloxacin (5 �g), ceftriax-ne (30 �g) and nalidixic acid (10 �g). MICs of five com-only used antimicrobials in paediatrics (amoxicillin, co-

rimoxazole, gentamicin, cefotaxime and ciprofloxacin) wereetermined using Etest strips (AB Biodisk, Solna, Sweden)ccording to the manufacturer’s instructions. Escherichiaoli ATCC 25922 (with known MIC) was used as a con-rol for the potency of antibiotics. Disk susceptibility testsnd MICs were interpreted according to the guidelines pro-ided by the National Committee for Clinical Laboratorytandards [13].

.4. Statistical analysis

Proportions of resistance in NTS were compared usinghe χ2 for trend by year or the Fisher’s exact test. Continuousariables were compared using the t-test or Mann–Whitney-test.

. Results

.1. Bacterial isolates

The majority of the 336 NTS studied were either S.nterica serovar Typhimurium (114; 33.9%) or S. entericaerovar Enteritidis (128; 38.1%); the proportions of these

wo serotypes did not differ significantly throughout thetudy period. The remaining 94 isolates (28%) consistedf 16% Salmonella Newport, 14% Salmonella Braenderup,1% Salmonella Derby, 11% Salmonella Colindale, 6%

168 S. Kariuki et al. / International Journal of Antimicrobial Agents 28 (2006) 166–171

Table 1Pulsed-field gel electrophoresis (PFGE) patterns produced by XbaI-digested DNA from 242 non-typhoidal Salmonella from blood cultures of children,1994–2005

Serotype PFGE group No. (%) in each PFGE group P-valuea

1994–1997 1998–2001 2002–2005

S. Typhimurium 1 36 (70.6) 37 (92.5) 16 (69.6)2 15 (29.4) 3 (7.5) 7 (30.4) 0.017

S. Enteritidis 1 40 (74.1) 48 (90.6) 13 (61.9)

SNh

3

SDp((a2t<Datpta(P

eStttpgdttmaditePo

F5sbS

2 5 (9.3)3 9 (16.7)

a Fisher’s exact test.

almonella Havana, 5% Salmonella Montevideo and 37%TS that were not serotypeable with the array of antiseraeld in our laboratory.

.2. Genotypes of NTS

PFGE was performed for all 114 S. Typhimurium and 128. Enteritidis using both XbaI- and SpeI-digested genomicNA. Salmonella Typhimurium isolates from the threehases of the study produced two patterns: one main patternPFGE group 1) consisting of a large proportion of isolates89; 78.1%) with PFGE bandings that were indistinguishable;nd one minor pattern (PFGE group 2) whose isolates (25;1.9%) produced banding patterns that differed by no morehan three bands (these differences occurred only for bands150 kb) (Table 1; Fig. 1). Using XbaI- and SpeI-digestedNA, the main PFGE pattern (group 1) consisted of 11 bands

nd 14 bands, respectively, between 50 kb and 600 kb. Fur-her analysis of the PFGE banding patterns across the threehases of the study did not produce any significant devia-

ion from the main PFGE digest patterns for the 1994–1997nd 2002–2005 phases of the study. However, the majority92.5%) of strains in the 1998–2001 study phase were inFGE group 1.

3

l

ig. 1. Restriction endonuclease fragment patterns of XbaI-digested DNA from re0 kb DNA molecular size standard; lanes 1–5, representative S. Typhimurium fromtrains in pulsed-field gel electrophoresis (PFGE) group 1; lanes 2–4 represent stlood cultures from 1998–2001 (lanes 6 and 7 represent strains in PFGE group 2; l. Typhimurium from blood cultures from 2002–2005 (lanes 13, 14 and 18 represen

5 (9.4) 6 (28.6)– 2 (9.5) 0.002

Salmonella Enteritidis serotype distribution in the differ-nt PFGE groups is shown in Table 1. Using both XbaI- andpeI-digested genomic DNA from S. Enteritidis, there werehree PFGE patterns for isolates from the three phases ofhe study. The commonest one (PFGE group 1) containedhe highest proportion of NTS isolates (101; 78.9%) and thisroduced PFGE patterns that were indistinguishable. PFGEroup 2 contained 16 isolates (12.5%) and produced PFGEigest patterns that differed from the main one by no morehan two bands. PFGE group 3 contained 11 isolates (8.6%)hat produced PFGE digest patterns that differed from the

ain one by three to five bands. Comparing S. Enteritidiscross the three phases of study, the 1998–2001 phase pro-uced the highest proportion of S. Enteritidis strains with anndistinguishable PFGE pattern (48; 90.6%) whilst the rest ofhe isolates (9.4%) were in PFGE group 2 (P < 0.01). How-ver, the distribution of S. Enteritidis strains within the threeFGE groups did not differ significantly for the two phasesf study 1994–1997 and 2002–2005.

.3. Antimicrobial susceptibility of bacterial isolates

There was a clear trend in reduction in the preva-ence of resistance to commonly used antimicrobials.

presentative Salmonella enterica serovar Typhimurium isolates. Lane M,blood cultures from children obtained in 1994–1997 (lanes 1 and 5 representrains in PFGE group 2); lanes 6–12, representative S. Typhimurium fromanes 8–12 represent strains in PFGE group 1); lanes 13–18, representativet strains in PFGE group 1; lanes 15–17 represent strains in PFGE group 2).

l of Ant

Tcg(

iteamat

tcTbi

tuasgt

4

bpmtfhctoecratu

oT

TM

A

A

C

C

C

G

M

S. Kariuki et al. / International Journa

he reduction was most remarkable for amoxicillin ando-trimoxazole (P < 0.001) but was also significant forentamicin (P = 0.005) and chloramphenicol (P = 0.05)Fig. 2).

For the period 1994–1997, 69.2% and 68.4% of NTS stud-ed were resistant to amoxicillin and co-trimoxazole, respec-ively, with corresponding high MIC values (Table 2). How-ver, all isolates tested were fully susceptible to cefotaximend ciprofloxacin, which are used mainly as second-line treat-ent for invasive bacterial infections both in children and

dults. In addition, only 12% of NTS isolates were resistanto gentamicin.

For the period 1998–2001, the proportion of NTS resistanto the two commonly used antimicrobials, amoxicillin ando-trimoxazole, declined to 43.5% and 36.2%, respectively.here was also a commensurate reduction in MIC values foroth antimicrobials (Table 2). Resistance levels for gentam-cin also fell to 7.2%.

For the period 2002–2005, there was a further reduc-ion in the prevalence of resistance to the two commonlysed antimicrobials, amoxicillin and co-trimoxazole, to 11%

nd 13%, respectively, and this was coupled with a corre-ponding reduction in MICs for the two antimicrobials. Forentamicin there was a further reduction of resistance levelso 4%.

Nath

able 2inimum inhibitory concentrations (MICs) and prevalence of resistance for antimi

ntimicrobial MIC characteristic MIC (�g/mL)

1994–1997 (n = 1

moxicillin Range 0.75–256Mode 256MIC50 256MIC90 256% Resistant 69.2

o-trimoxazole Range 0.05–32Mode 32MIC50 32MIC90 32% Resistant 68.4

iprofloxacin Range 0.01–0.05Mode 0.01MIC50 0.01MIC90 0.02% Resistant 0

efotaxime Range 0.03–1.5Mode 0.06MIC50 0.13MIC90 0.38% Resistant 0

entamicin Range 0.13–16Mode 8MIC50 0.38MIC90 12% Resistant 12

IC50 and MIC90, minimum inhibitory concentration required to inhibit the growt

imicrobial Agents 28 (2006) 166–171 169

. Discussion

In Africa, NTS remains an important cause of childhoodacteraemia and is responsible for high mortality rates ifrompt diagnosis and appropriate antibiotic treatment is notade available. For instance, in Blantyre, Malawi, NTS was

he most common blood culture isolate (40%), with a caseatality rate of 24% [4]. In sub-Saharan Africa, most publicospitals lack adequate laboratory facilities to diagnose spe-ific causes of sepsis and only perform antimicrobial suscep-ibility tests later on. Furthermore, most deaths from sepsisccur within 48 h, before culture results would be availableven where facilities exist. Thus, physicians rely on empiri-al treatment using commonly available antimicrobials. Ouresults suggest that in coastal Kenya, the commonly usedntimicrobials are still effective for treatment of NTS at aime when the general trend for resistance in NTS from therban population has been upward.

Over the 12-year study period, the two main serotypesf NTS causing bacteraemia in children in Kilifi (S.yphimurium and S. Enteritidis) accounted for 72% of allTS and this remained stable and almost equally distributed

cross the three phases of the study. PFGE analysis of thesewo main serotypes showed that, within each serotype, strainsad minimal genetic diversity and that the strains appeared

crobials tested during the 12-year study period

33) 1998–2001 (n = 138) 2002–2005 (n = 65)

0.5–256 0.5 to >256256 2

1 2256 443.5 11

0.03–32 0.05–3232 0.20

0.13 0.2032 3236.2 13

0.01–0.09 0.01–0.060.01 0.010.01 0.010.02 0.020 0

0.06–4 0.02–0.40.06 0.050.09 0.050.38 0.060 0

0.1–24 0.1–80.38 0.20.38 0.258 27.2 4

h of 50% and 90% of the organisms, respectively.

170 S. Kariuki et al. / International Journal of Ant

FrcP

totPrtd

csclrmgeftTsiroanmo

apoiac4

5hehibt

ocadmcopaTcNhwtta

ttaSoanpoHcbaci1afhuttpq

b

ig. 2. Trends in resistance during the 12-year study period: chlo-amphenicol (χ2 = 3.794; P = 0.05); gentamicin (χ2 = 7.958; P = 0.005);o-trimoxazole (χ2 = 16.358; P < 0.001); and amoxicillin (χ2 = 20.977;< 0.001). NTS, non-typhoidal Salmonella.

o be clonally related. It appears that this limited diversityf strains was stably maintained over the 12-year period ofhe study: for S. Typhimurium, 78.1% of strains producedFGE digest patterns that were indistinguishable whilst theest (21.9%) produced digest patterns that differed by no morehan three bands; for S. Enteritidis, 78.9% of the strains pro-uced an identical PFGE digest pattern.

All the NTS remained fully susceptible to cefotaxime andiprofloxacin. In Kilifi, cephalosporins are used mainly asecond-line treatment for invasive bacterial infections both inhildren and adults in this setting. Our analysis of resistanceevels among NTS has revealed a remarkable decreasingate of prevalence of resistance especially to the most com-only available antimicrobials (amoxicillin, co-trimoxazole,

entamicin and chloramphenicol) during 1994–2005. In anarlier study of the susceptibility of Gram-negative bacteriarom the same population, Bejon et al. [14] initially observedhe trend of decreasing resistance between 1998 and 2001.he decrease was attributed to a change in blood cultureampling policy, from selective to all children admitted, thusncluding those with milder disease who are less likely to haveeceived antibiotics before presentation. However, the trendf decreasing resistance levels in NTS has continued steadilyfter 2001, suggesting that changes in sampling criteria areot primarily responsible for the effect observed. Further-ore, from 1998, admission rates and laboratory processing

f blood culture samples have not changed.In contrast, previous studies carried out on NTS from

dult cases of bacteraemia treated at three main urban hos-itals in Kenya showed a clear trend of increasing levelsf resistance to the same group of antimicrobials [15]. For

nstance, the level of resistance for some of the commonlyvailable antimicrobials, including ampicillin, streptomycin,o-trimoxazole and chloramphenicol, rose from 48%, 49%,6% and 26%, respectively, in 1993 to 65%, 67%, 68% and

siaa

imicrobial Agents 28 (2006) 166–171

0%, respectively, in 2003 (P < 0.01), with correspondinglyigh MICs for these antimicrobials. Higher resistance lev-ls in the urban population may be due to higher usage inealth clinics for the treatment of diarrhoeal and respiratorynfections and misuse of commonly available antimicrobialsought over the counter for self-treatment without prescrip-ion [15].

One limitation of our study was that we were unable tobtain data on levels of usage of commonly available antimi-robials in the rural community and so may not with certaintyttribute this continued decline in levels of resistance toecreased use of amoxicillin and co-trimoxazole in the com-unity health clinics and/or less availability for sale over the

ounter without prescription. Other limitations were the lackf detailed data on human immunodeficiency virus amongaediatric admissions and on changes in the use of antifolatentimalarials that could affect resistance to co-trimoxazole.hese two factors could have independent effects on antimi-robial resistance of NTS. In addition, the smaller number ofTS analysed for the 2002–2005 phase of the study may alsoave introduced some element of sampling error. However,e used the same method of Etest to determine MICs over

he whole study period and there is a clear trend of reduc-ion in prevalence of resistance to the commonly availablentimicrobials.

A few studies in other parts of the world have also shownhat bacterial strains may show a reduced resistance pheno-ype after a period of withdrawal or cycling of particularntibiotics. For instance, in Bangladesh, multidrug-resistantalmonella Typhi strains formed 46.5%, 29.7% and 17.8%f all NTS in 1994, 1995 and 1996, respectively (P < 0.001),nd resistance levels to chloramphenicol also decreased sig-ificantly. This was attributed to a shift from using chloram-henicol in treating enteric fever following the availabilityf ceftriaxone and ciprofloxacin in 1991 [16]. In Taiwan,sueh [17] also observed a 16% decline in rates of peni-

illin resistance among Streptococcus pneumoniae isolatesetween 1998 and 1999 (25%) and 2001 (9%), and this wasssociated with a 46% decrease in total penicillin and otherephalosporin usage in 2001 compared with 1999. Finally,n the USA, Heilmann et al. [18] observed that between994 and 2003 the prevalence of �-lactamase productionnd resistance to �-lactams among Haemophilus influenzaerom patients with respiratory tract infections appeared toave decreased, which they attributed to a shift away fromsing amoxicillin and less potent oral cephalosporins in thereatment of community-acquired respiratory tract infectionsoward use of amoxicillin/clavulanic acid, macrolides, moreotent advanced generation oral cephalosporins and fluoro-uinolones.

In conclusion, our study of NTS from children withacteraemia in rural Kilifi district over a 12-year period

howed minimal diversity in strain genotypes and a signif-cant reduction in resistance levels to commonly availablentimicrobials. It is noteworthy that in health clinics in suchrural situation, the majority of cases of even severe invasive

l of An

bibop

A

tAdKJuT

R

[

[

[

[

[

[

[

[

S. Kariuki et al. / International Journa

acterial infections are often treated empirically owing tonadequate facilities for diagnosis and antimicrobial suscepti-ility testing. To determine whether this decline in prevalencef antimicrobial resistance is a permanent feature will requireeriodic and sustained surveillance and monitoring.

cknowledgments

We thank the Director, Kenya Medical Research Insti-ute, for permission to publish this work. We are grateful tonthony Scott for help with data analysis; Rhoda Njagi forata management; Joyce Mwituria, Jane Muyodi, Dorothyagendo and Agnes Munyalo for laboratory analysis; and

ames Waithaka and Charles Lang’at for assistance in follow-p field studies. S.K. is supported by a Wellcome Trustraining Fellowship.

eferences

[1] Pegues DA, Miller SI. Salmonellosis, including typhoid fever. CurrOpin Infect Dis 1994;7:616–23.

[2] Sirinavin S, Jayanetra P, Thakkinstian A. Clinical and prognosticcategorization of extraintestinal nontyphoidal Salmonella infectionsin infants and children. Clin Infect Dis 1999;29:1151–6.

[3] Yang YJ, Huang MC, Wang SM, Wu JJ, Cheng CP, Liu CC.Analysis of risk factors for bacteremia in children with nonty-phoidal Salmonella gastroenteritis. Eur J Clin Microbiol Infect Dis2002;21:290–3.

[4] Graham SM, Molyneux EM, Walsh AL, Cheesbrough JS, MolyneuxME, Hart CA. Nontyphoidal Salmonella infections of children intropical Africa. Pediatr Infect Dis J 2000;19:1189–96.

[5] Graham SM. Salmonellosis in children in developing and developed

countries and populations. Curr Opin Infect Dis 2002;15:507–12.

[6] Berkley JA, Lowe BS, Mwangi I, et al. Bacteremia among childrenadmitted to a rural hospital in Kenya. N Engl J Med 2005;352:39–47.

[7] Mwangi I, Berkley J, Lowe B, Peshu N, Marsh K, Newton CR.Acute bacterial meningitis in children admitted to a rural Kenyan

[

timicrobial Agents 28 (2006) 166–171 171

hospital: increasing antibiotic resistance and outcome. Pediatr InfectDis J 2002;21:1042–8.

[8] Su LH, Wu TL, Chia JH, Chu C, Kuo AJ, Chiu CH. Increasing cef-triaxone resistance in Salmonella isolates from a university hospitalin Taiwan. J Antimicrob Chemother 2005;55:846–52.

[9] Choi SH, Woo JH, Lee JE, et al. Increasing incidence of quinoloneresistance in human non-typhoid Salmonella enterica isolates inKorea and mechanisms involved in quinolone resistance. J Antimi-crob Chemother 2005;56:1111–4.

10] World Health Organization. Management of the child with a seriousinfection or severe malnutrition: guidelines for care at the first-referral level in developing countries. Geneva, Switzerland: WHO;2000.

11] Kariuki S, Oundo JO, Muyodi J, Lowe B, Threlfall EJ, HartCA. Genotypes of multidrug-resistant Salmonella enterica serotypetyphimurium from two regions of Kenya. FEMS Immunol MedMicrobiol 2000;29:9–13.

12] Tenover FC, Arbeit RD, Goering RV, et al. Interpreting chromo-somal DNA restriction patterns produced by pulsed-field gel elec-trophoresis: criteria for bacterial strain typing. J Clin Microbiol1995;33:2233–9.

13] National Committee for Clinical Laboratory Standards. Performancestandards for antimicrobial susceptibility testing; twelfth infor-mational supplement. Approved standard M100-S12. Wayne, PA:NCCLS; 2002.

14] Bejon P, Mwangi I, Ngetsa C, et al. Invasive Gram-negative bacilliare frequently resistant to standard antibiotics for children admittedto hospital in Kilifi, Kenya. J Antimicrob Chemother 2005;56:232–5.

15] Kariuki S, Revathi G, Kariuki N, et al. Increasing prevalence ofmultidrug-resistant non-typhoidal salmonellae, Kenya, 1994–2003.Int J Antimicrob Agents 2005;25:38–43.

16] Saha SK, Saha S, Ruhulamin M, Hanif M, Islam M. Decreasingtrend of multiresistant Salmonella typhi in Bangladesh. J AntimicrobChemother 1997;39:554–6.

17] Hsueh PR. Decreasing rates of resistance to penicillin, but noterythromycin, in Streptococcus pneumoniae after introduction of apolicy to restrict antibiotic usage in Taiwan. Clin Microbiol Infect

2005;11:925–7.

18] Heilmann KP, Rice CL, Miller AL, et al. Decreasing prevalenceof beta-lactamase production among respiratory tract isolates ofHaemophilus influenzae in the United States. Antimicrob AgentsChemother 2005;49:2561–4.