Paediatric antimicrobial use at a South African hospital

Paediatric

antimicrobial

use

at

a

South

African

hospital

L.R.

Koopmans

,

H.

Finlayson

,

A.

Whitelaw

,

E.H.

Decloedt

,

A.

Dramowski

*

a

UndergraduateResearchElectiveProgramme,DepartmentofPaediatricsandChildHealth,StellenboschUniversity,CapeTown,SouthAfrica

b

RadboudUniversity,RadboudUMC,TheNetherlands

c

DepartmentofPaediatricsandChildHealth,DivisionofPaediatricInfectiousDiseases,FacultyofMedicineandHealthSciences,StellenboschUniversity,Cape Town,SouthAfrica

d_{DepartmentofMedicalMicrobiology,FacultyofMedicineandHealthSciences,StellenboschUniversity,CapeTown,SouthAfrica} e_{NationalHealthLaboratoryServices,TygerbergHospital,CapeTown,SouthAfrica}

f

DepartmentofMedicine,DivisionofClinicalPharmacology,FacultyofMedicineandHealthSciences,StellenboschUniversity,CapeTown,SouthAfrica

ARTICLE INFO

Articlehistory: Received16March2018

Receivedinrevisedform24May2018 Accepted29May2018

CorrespondingEditor:EskildPetersen, Aar-hus,Denmark Keywords: Antibiotic Antimicrobial Paediatric Antibioticstewardship Surveillance Prescription Pharmacy ABSTRACT

Background:DataonantimicrobialuseamonghospitalizedchildreninAfricaareverylimitedduetothe absenceofelectronicprescriptiontracking.

Methods:Thisstudyevaluatedantimicrobialconsumptionrates,theantimicrobialspectrumused,and theindicationsfortherapyonapaediatricwardandinthepaediatricintensivecareunit(PICU)at TygerbergHospital,CapeTown,SouthAfrica.Antimicrobialprescriptionandpatientdemographicdata werecollectedprospectivelyfromMay10,2015toNovember11,2015.Forthesameperiod,dataon antimicrobialsdispensedandcostswereextractedfromthepharmacyelectronicmedicinemanagement system. The volume of antimicrobials dispensed (dispensing data) was compared with observed antimicrobialuse(prescriptiondata).

Results:Ofthe703patientsadmitted,415/451(92%)paediatricwardadmissionsand233/252(92%)PICU admissionsreceived1antimicrobials.Ontheward,89%ofprescriptionswereforcommunity-acquired infections;29%ofPICUantimicrobialswereprescribedforhealthcare-associatedinfections.Ampicillin andthird-generationcephalosporinswerethemostcommonlyprescribedagents.Antimicrobialcosts were67541SouthAfricanRand(ZAR)(5680UnitedStatesDollars(USD))onthewardand210484ZAR (17702 USD) in the PICU. Ertapenem and meropenem were the single largest contributors to antimicrobial costson theward (43%) andPICU(30%), respectively.The volumeof antimicrobials dispensedbythepharmacy(dispensingdata)differedconsiderablyfromobservedantimicrobialuse (prescriptiondata).

Conclusions:Highratesofantimicrobialconsumptionweredocumented.Community-acquiredinfections were the main indication for prescription. Although pharmacy dispensing data did not closely approximateobserveduse,thisrepresentsapromisingmethodforantimicrobialusagetrackinginthe future.

©2018TheAuthor(s).PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases. ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(

http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Antimicrobial overuse is an important contributor to the developmentofantimicrobialresistanceworldwide.Antimicrobial consumption is particularly high among paediatric inpatients owing to a predominance of infectious pathologies (both community-andhealthcare-associatedinfection)andnon-speci_fic disease presentations, with difficulty in excluding bacterial infections. Few studies have assessed the appropriateness of

antimicrobial use in children, but in high-income settings, antimicrobialprescribingerrorsoccurfrequently,including incor-rectdosing,inappropriateantimicrobialchoice,noindicationfor therapy,inadequatetreatmentduration,andinappropriateroute of administration (Blinova et al., 2013; Newland et al., 2012; Kreitmeyretal.,2017).

A recent point prevalence study (Antibiotic Resistance and Prescribing in European Children – ARPEC) surveyed 17693 paediatric inpatients in 41 countries, including six African hospitals (Versporten et al., 2013). Bacterial lower respiratory tractinfectionswerethemostcommonindicationfor antimicro-bialprescriptionworldwide.AtthesixAfricanhospitalsstudied, gentamicinandceftriaxonewerethemostcommonlyprescribed

*Correspondingauthor.

E-mailaddress:dramowski@sun.ac.za(A.Dramowski).

https://doi.org/10.1016/j.ijid.2018.05.020

1201-9712/©2018TheAuthor(s).PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

ContentslistsavailableatScienceDirect

International

Journal

of

Infectious

Diseases

antimicrobials, contributing 16%and 14%, respectively, of total antimicrobialusage.InanEthiopianpaediatricinpatientsetting, gentamicin(26%)andceftriaxone(44%)werealsoidentifiedas the most commonly prescribed antimicrobials (Feleke et al., 2013).IntheARPECstudy,37%ofallpaediatricinpatientshadone ormoreantimicrobialsprescribed;thehighestantimicrobialuse rates(61%)weredocumentedinpaediatricintensivecareunits (PICU).Otherinternationalstudieshaveconfirmedhigh antimi-crobialusagerates(rangingfrom32%to70%)(Grohskopfetal., 2005;Amadeoet al.,2010; Gandraet al., 2017;DeLuca etal., 2016), with the highest paediatric inpatient antimicrobial consumptionratesreportedfromAfricancountries:Mozambique (98%), Botswana (72%), Ethiopia (98%), and Ghana (69%) (Monteiro et al., 2017; Labi et al., 2018; Fisher et al., 2009; AlemnewandAtnafie,2015).

A specific challengeto monitoring and reporting paediatric antimicrobialconsumptiondataisthelackofastandardizedusage definition.Among adult patients, consumption patterns can be easily compared using the defined daily dose (DDD) for each antimicrobialagent(totalgramsofthedrugdividedbythenumber ofgramsinanaverageadultdailydose)(WHO,2018).Thismethod cannot be applied accurately to paediatric populations as antimicrobialdosageinchildrenisbasedonbodyweightorbody surfacearea.

Accurateand ongoingmeasurement of local antimicrobial consumptionpatterns helps stewardship teams set program-matictargetsandassesstheimpactofinterventions.Many high-income countries use electronic laboratory and pharmacy surveillance systems to measure usage trends and generate real-time/early warning systems for infectious disease out-breaks,supportingantimicrobialstewardshipefforts.However, in resource-limited healthcare settings and particularly in Africa, surveillance of antimicrobial use through electronic prescribing is generally unavailable (Rattanaumpawan et al., 2017).Furthermore,thelackofelectronicprescriptiontracking means that any antimicrobial consumption data must be collectedmanually,whichislabour-intensiveandmoreprone to error. The lack of tools for antimicrobial consumption monitoring hampers the publication of data on paediatric antimicrobial use from low- and middle-income countries (LMIC)(Dillonetal.,2014).

Althoughelectronic prescribingis notavailable, electronic pharmacystock managementtools are widely used in public sectorSouthAfrican healthcarefacilities.Thepotential useof electronic dispensing data for tracking antimicrobial usage trendsisappealing,as itavoidstheneedforlabour-intensive prescription audits.However, pharmacy dispensing datamay notalways accuratelyreflectusagedata;forexample, in low-resourceneonatal/paediatricsettings,thesharingofmulti-dose antimicrobialvials betweenpatientsis acommonpractice. In suchcases,dispensingdatawouldonlyre_flecttheinitialpatient to whom the antimicrobial was dispensed and not the subsequentpatients.

GiventhepaucityofdataonantimicrobialuseinAfrica,this studywasperformedtoevaluateantimicrobialconsumption on two paediatric wards, comparing prescription chart auditdata withpharmacydispensingdataatTygerbergHospital,CapeTown, SouthAfrica.

Methods Studysetting

TygerbergHospitalinCapeTown,SouthAfricaisalarge 1384-bed academic complex,including 300 neonatal/paediatric beds and17000neonatal/paediatric(0–14years)admissionsannually.

Thereare13children’swards:medicalgeneralistwards,medical specialistwards,surgicalwards,neonatalwards,andtwointensive careunits–neonatalandpaediatric(PICU).Community-acquired infectious (CAI) diseases like human immunodeficiency virus (HIV),tuberculosis(TB),respiratorytractinfections,and gastroen-teritisarecommonindicationsforpaediatrichospitalization.The treatment ofpaediatrichealthcare-associatedinfections(HAI)is also a major contributor to antimicrobial use on the wards (Dramowski et al., 2016). The hospital implemented a formal antimicrobial stewardship programme in 2014 with annual antimicrobialusepointprevalencesurveys,weeklyantimicrobial stewardshipwardrounds,andadedicatedantimicrobial prescrip-tion chart. The hospital pharmacy has utilized an electronic medicine management system (JAC Medicines Management software)sincetheyear2000;however,comprehensivedataon antimicrobial consumption ratesby ward are notdisseminated routinely. The JAC software system manages and streamlines procurement, stock keeping, dispensing, and distribution of pharmaceuticalswithinthepharmacy,butthehospitaldoesnot currentlyhaveaccesstotheelectronicprescribing JACsoftware module.

Studydesign

FromMay10,2015toNovember11,2015,datafrominpatient records and antimicrobial prescriptions were collected pro-spectively on two wards (the PICU and a general paediatric ward) for all children admitted for >48h during a study to establish the incidence of paediatric HAI (Dramowski et al., 2016).Thisdataset(6-monthantimicrobialprescriptionaudit) includeddrugname,dose,duration,andindicationtype (CAI, HAI, or prophylaxis). Antiretrovirals, anti-TB drugs, and co-trimoxazole prophylaxis for Pneumocystis jirovecii were not includedintheaudit.Forthesame6-monthperiodandthesame two wards, datawere extracted from the pharmacy electronic medicinemanagementsystem (JAC),includingdrugname,total units dispensed, total consumption (in milligrams, grams, or international units),and cost(inSouthAfricanRand).Observed antimicrobialusewasdescribedusingtheprospectivelycollected antimicrobialprescriptionauditdata.Antimicrobialcostdatawere derived from the pharmacy dispensing data. The two datasets were compared to determinethe variance between audit data (observeduse)anddispensingdataandthefeasibilityofusingthe pharmacy dispensing data for future surveillance of paediatric antimicrobialuse.

Studydefinitions

CAIincludedallinfectiousdiseasesthatwereacquiredprior to the current hospitalization episode. HAI included any infectionthat wasneitherpresentnor incubatingat the time of hospital admission (National Healthcare Safety Network, 2013). Antimicrobial prophylaxis included the prescription of antimicrobialsforthepurposeofpreventinginpatientsurgical ormedicalinfections(otherthanco-trimoxazoleprophylaxisfor Pneumocystisjirovecii,isoniazidprophylactictherapy,and antire-troviral prophylaxis for the prevention of mother-to-child transmission of HIV). ‘Patient-days’ was defined as thesum of allpatientlengthsofstayperwardpermonth.Theantimicrobial consumptionorutilizationratewasdefinedasthetotalproportion of patients over the study period who received one or more antimicrobial drugs during their admission episode. Days of therapy(DOT)wascalculatedforeachantimicrobialagentasthe sumofallpatientusageofthatdruginthetotalnumberofdays received (Tadesse et al., 2017). For calculation of antimicrobial costs,thecurrencyconversionrateforSouthAfricanRand(ZAR)

andUnitedStatesDollars(USD)thatwasapplicableinMay2015 wasused(1USD=11.89ZAR).

Datahandlingandanalysis,andethicalapproval

Thefollowingindiceswerecalculatedforeachwardusingthe prescriptionauditdataonly(asthenecessaryvariableswerenot available from the dispensing data): (1) the antimicrobial prescriptionprevalencerate;(2)themeannumberof antimicro-bials per patient admission episode; (3) the indication/s for antimicrobial/s (CAI, HAI, prophylaxis); (4) the antimicrobial spectrumprescribed(antibacterial,antifungal,antiviral,andthe proportionalcontributionofeachdrugtotheoverallantimicrobial usage); (5) the length of hospital stay versus number of antimicrobials prescribed. Antimicrobial costs were calculated from the dispensing data. Audit data were compared with dispensing data to calculate the variance in usage between datasets.Insomeinstances,childrenhadmorethanoneadmission episode to the selected wards during the study period; these admission episodes were included using the patient’s current weight,age,andprescriptionindication.

Continuous and categorical data were analyzed using the Studentt-testandFisher’sexacttest/Chi-squaretest,as appropri-ate.Pearson’scorrelationcoefficientwasusedtoinvestigatethe relationshipbetweenlengthofhospitalstayandthenumberof antimicrobials prescribed. A p-value of <0.05 was considered statistically significant. Stata statistical software version 13.1 (StataCorp,USA)wasusedforthedataanalysis.

Ethicalapprovalandawaiverofindividualinformedconsent wasobtainedfromtheHumanHealthResearchEthicsCommittee ofStellenboschUniversity(Ref.No.S13/09/171).

Results

Atotalof703patientswereadmittedduringthestudyperiod: 451tothegeneralpaediatricwardand252tothePICU(Table1).Of patientsadmittedtotheward,415/451(92%)wereprescribedone ormoreantimicrobialsduringtheirhospitalstay;anaverageof2.2 antimicrobialswereprescribedperadmissionepisode(range0–8 antimicrobials). Overall on the ward, 904 antimicrobials were prescribed,generating 4079DOTand ausagerateof1137DOT/ 1000patient-days.AmongpatientsadmittedtothePICU,233/252 (92%)wereprescribedoneormoreantimicrobialdrugswithan averageof2.8antimicrobialsperpatient(range0_–11).Overall,662 antimicrobialswereprescribed,generating3810DOTandausage rateof1323DOT/1000patient-days.Youngerpatients(neonates andinfants)hadsignificantlyhigherantimicrobialusageratesthan olderchildren,bothonthewardandinthePICU.Patientsinthe PICUweresignificantlymorelikelytobeprescribedan antimicro-bialforHAIthanpatientsontheward(p<0.001).Asthenumberof antimicrobialsprescribedperpatientincreased,themeanduration of stay increased significantly for the population (Pearson correlation coefficient 0.360 (moderate correlation), p<0.001) (Figure1a).ForbothwardandPICUcombined(n=703),patients whodied(n=23)receivedsignificantlymoreantimicrobialsthan thosewhosurvived(medianof5vs.2antimicrobials,respectively) (Figure1b).

Mostantimicrobialsweregivenparenterally(72%ontheward and 87% in the PICU). The third-generation cephalosporins (cefotaximeandceftriaxone)werethemostcommonlyprescribed intravenous(IV)antimicrobialsonthepaediatricward(833DOT) andinthePICU(436DOT).Ontheward,ampicillin(757DOT)and gentamicin(442DOT)werethesecondandthirdmostfrequently

Table1

Demographicsofthestudypopulation(n=703).

Paediatricward(n=451) PaediatricICU(n=252) 1antimicrobial prescription Noantimicrobial prescription p-Value 1antimicrobial prescription Noantimicrobial prescription p-Value

Proportionofpatientsprescribed1 antimicrobialdrug

415(92%) 36(8%) – 233(92%) 19(8%) – Totalnumberofantimicrobialsprescribed 904 NA – 662 NA – Meanantimicrobialsperpatient,n(range) 2.2(1–8) NA – 2.8(1–11) NA – Sex,male 233(56.1%) 22(61.1%) 0.564 130(55.8%) 12(63.1%) 0.534 Ageinmonths,median(IQR) 5.5(1.6–21.5) 27.4(6.1–86.4) <0.001 5.9(2.2–21.5) 27.9(10.0–47.5) 0.001

Agecategory <0.001 0.015

Neonate(<28days) 57(13.7%) 0(0%) 25(10.7%) 0(0%) Infant(29–364days) 212(51.1%) 13(36.1%) 125(53.7%) 6(31.6%) Toddler(1–5years) 104(25.1%) 12(33.3%) 59(25.3%) 11(57.9%) Child(>5years) 42(10.1%) 11(30.6%) 24(10.3%) 2(10.5%)

Weightinkilograms,median(IQR) 5.7(3.7–9.4) 9.4(5.8–15.7) <0.001 6.5(3.6–10.5) 9.8(7.3–13.6) 0.036

HIVprevalence 0.815 0.05

HIV-infected 25(6.0%) 2(5.5%) 24(10.3%) 0(0%) HIV-exposeduninfected 77(18.6%) 5(13.9%) 31(13.3%) 2(10.5%) HIV-negative 293(70.6%) 28(77.8%) 164(70.4%) 13(68.4%) HIVunknown 20(4.8%) 1(2.8%) 14(6.0%) 4(21.1%)

Lengthofstayindays,mean(SD) 6(4–9) 6(3–9) 0.437 9(5–13) 3(2–4) <0.001

Indicationforantimicrobial/s – –

Community-acquiredinfection 809/904(89%) NA 464/662(70%) NA Healthcare-associatedinfection 94/904(10%) 191/662(29%) Prophylaxisa 1/904(0%) 7/662(1%) Outcome 0.609 0.183 Dischargedortransferred 412(99.3%) 36(100%) 213(91.4%) 19(100%) Died 3(0.7%) 0(0%) 20(8.6%) 0(0%)

ICU,intensivecareunit;NA,notapplicable;IQR,interquartilerange;SD,standarddeviation;HIV,humanimmunodeficiencyvirus.

a

Prophylaxisincludedmedicalandsurgicalprophylaxisotherthanco-trimoxazoleprophylaxisforPneumocystisjirovecii,isoniazidprophylactictherapy,andantiretroviral prophylaxisforpreventionofmother-to-childtransmissionofHIV.

prescribedIVantimicrobials.InthePICU,meropenem(422DOT) and ganciclovir (383 DOT) were the second and third most frequentlyprescribedIVantimicrobials(Figure2a,b).Commonly usedoralantimicrobialsonthewardwereamoxicillin(489DOT), amoxicillin–clavulanicacid(271DOT),andpenicillinVK(81DOT). InthePICU,co-trimoxazolewasthemostcommonlyprescribed oralantimicrobial(141DOT),followedbyamoxicillin–clavulanic acid(129DOT)anderythromycin(89DOT)(Figure2c,d).

CAIwere the main indication for antimicrobial prescription (89%ontheward,70%inthePICU;p<0.001)withampicillinand third-generationcephalosporinsbeingthemostcommonagents prescribedforCAI(Figure3a,c).Ertapenemandmeropenemwere the mostcommonly prescribedantimicrobials for HAIs on the wardandinthePICU(Figure3b,d).Prophylacticantimicrobials constitutedaverysmallpercentageofoverallusageontheward andinthePICU(0.1%and1.1%,respectively).

Ofthedifferentantimicrobialclasses,antibacterialagentswere mostcommonlyused(94%onthewardand81%inthePICU),with antifungals and antivirals making up a small proportion of antimicrobials used. However, in terms of days of therapy, ganciclovir, aciclovir, and valganciclovir usage was substantial, beingthe4th,11th,and14thoveralllargestcontributorstoDOTin the PICU. There was also a greater diversity of antimicrobials prescribedinthePICUthanontheward.

Antimicrobialcostscalculatedfromthepharmacydispensing data estimated a total antimicrobial spend on the general paediatricwardof67541ZAR(USD5680)intotal,equatingto 162 ZAR (USD 13.69) per infected patient. In the PICU, antimicrobial prescriptionstotalled210484ZAR (USD17702), and903ZAR(USD75.98)perinfectedpatient.Ertapenemand meropenemwerethesinglelargestcontributortoantimicrobial costsontheward(43%of totalcosts)andinthePICU(30%of

totalcosts). Thefive antimicrobials with the highestcost are showninTable2.

The most commonly prescribed IV antimicrobials from the prescriptionauditswerecomparedwiththevolumeofthesame antimicrobialagents dispensedaccordingtothepharmacydata (Table3).Forthegeneralpaediatricward,withtheexceptionof amoxicillin–clavulanic acid and ertapenem, consumption exceeded volumes dispensed by up to 2.8-fold. Conversely in thePICU,alargervolumeofantimicrobialswasdispensedthanwas actuallyconsumed.

Discussion

These datarepresentoneof thefirstoverviews ofpaediatric antimicrobialuseata SouthAfricanhospital.Veryhighratesof antimicrobialuseweredocumentedbothonthewardandinthe PICU, exceeding rates reported from most other paediatric inpatient settings worldwide (Feleke et al., 2013; Grohskopf etal., 2005; Amadeo et al.,2010; Gandraet al., 2017;De Luca etal.,2016).IncomparisontoaGermanstudyofantimicrobialuse ongeneralpaediatricwards(reportingDOTof483/1000 patient-days)(Kreitmeyretal.,2017),thegeneralwardinthepresentstudy utilized 1137 DOT/1000 patient-days. This substantially higher usagemaybeascribedtotheinclusionofantiviralsandantifungals inthisstudy,andthegreaterburdenofinfectiousdiseasesamong thepaediatricadmissions.However,theantimicrobialusagedata arenotdissimilartothosereportedinthefewpublishedstudies fromotherAfricanpaediatricwards(Monteiroetal.,2017;Labi etal.,2018;Fisheretal.,2009;AlemnewandAtnafie,2015).The presentstudydatasupportexistingevidencethatantimicrobial use on the continent is substantial, highlighting the need for greatersurveillanceofantimicrobialuseandtheimplementation ofantimicrobialstewardshipprogrammesinAfricanhospitals.

InthisTygerbergHospital cohort,it waspossibletoshowa significantassociationbetweenantimicrobialuseandyoungage

(every neonate had at least one antimicrobial prescribed), highlighting the dif_ficulty in excluding infectious diseases in theneonate/younginfant.Furthermore,anassociationbetween longer length of stay and greater number of antimicrobials prescribed wasdemonstrated, whichprobably reflects compli-catedclinicaldiseasecourseandthegreaterlikelihoodofHAIas the length of hospital stay increases. Similarly, a significant associationbetweenantimicrobialuseandoutcomewasshown, withPICUpatientswhoreceivedantimicrobials havinga crude mortalityrateof8.6%versusnodeathsamongpatientswhodid notreceiveantimicrobials.

IntheglobalARPECpointprevalencestudy,gentamicinand ceftriaxonewerethemostcommonlyprescribedantimicrobials amongpaediatricinpatients,whereasat thestudyinstitution, third-generation cephalosporins, ampicillin, and gentamicin were themost widelyprescribedantimicrobials. Forboththe ward and PICU, CAI was the most common indication for receiving an antimicrobial/s: third-generation cephalosporins (23%onthegeneralwardand19%inthePICU)andampicillin (25% on the general ward and 18% in the PICU) were most commonly used. In a Ghanaian teaching hospital, third-generation cephalosporins were commonly used for CAIs (28%)(Labietal.,2018).

For HAI, the study institution predominantly used carbape-nems,namelymeropenem(31%onthegeneralwardand34%inthe PICU)and ertapenem(39%onthegeneralwardand 19%in the PICU).Aspartofongoingantimicrobialstewardshipefforts, broad-spectrumantimicrobialuseforsuspectedHAIrequiresconsultant approval and a named-patient prescription to pharmacy for authorization. This high carbapenemusage mostlikely reflects the hospital guidelines in 2015 at the time of data collection. Subsequently,therecommendedempiricHAItreatmentforward patients has changed to piperacillin–tazobactam and amikacin (personalcommunication,DrHeatherFinlayson),whichhasledto an increase in consumption of these antibiotics. Clinicians are

Figure2.(a)Totaldaysofintravenous(IV)antimicrobialtherapyonthepaediatricward.(b)Totaldaysofintravenous(IV)antimicrobialtherapyinthepaediatricICU.(c)Total daysoforalantimicrobialtherapyonthepaediatricward.(d)TotaldaysoforalantimicrobialtherapyinthepaediatricICU.

encouraged (through the dedicated antimicrobial therapy pre-scriptionchart),tostoporde-escalatetheempiricbroad-spectrum antimicrobial therapy after 72hif the initial laboratory inves-tigations for sepsis are negative. HAI as an indication for antimicrobialprescriptionoccurredmorefrequentlyinthePICU, although this finding was not surprising given the severity of illness,useofindwellingdevices,and extendedhospitalstayof populationsinanICUsetting.

ProphylaxisasanindicationforantimicrobialuseinthePICU andonthewardwasrare,reflectingthesmallnumberofpaediatric surgicalpatientsincluded.Arecentantimicrobialpointprevalence survey in a teaching hospital in Ghana showed a different distribution of indications for antimicrobial therapy compared tothestudyinstitution:40%wereforCAI,21%forHAI,34% for surgicalprophylaxis,and5%formedicalprophylaxis(thisstudy includedbothadultandpaediatricmedicalandsurgicalwards).

Figure3.(a)Antimicrobialsusedforcommunity-acquiredinfectiononthepaediatricward(n=809).(b)Antimicrobialsusedforhealthcare-associatedinfectiononthe paediatricward(n=94).(c)Antimicrobialsusedforcommunity-acquiredinfectioninthepaediatricICU(n=464).(d)Antimicrobialsusedforhealthcare-associatedinfection inthepaediatricICU(n=191).

Thetotalcostofantimicrobialsdispenseddifferedsubstantially bywardoforigin:theantimicrobialcostperinfectedpatientwas five-foldhigherforPICUpatients(13.69vs.75.98USDperinfected patient). There are several potential explanations for the cost differential:PICUpatientshadhigherratesofparenteraldruguse, higherratesofHAI(withgreateruseofcarbapenems),extended lengths of stay, and greater use of antivirals. However, an additionalfactor contributingtothis cost calculationmayhave beenoverestimationofthevolumeofantimicrobialsdispensedby thepharmacy.Inhigh-incomecountries,ongoinganalysisofdrug utilizationratesandprescribingtrendshasbeenusedsuccessfully byhospitalantimicrobialstewardshipprogrammes(Reddyetal., 2015;AraujodaSilvaetal.,2017;Smithetal.,2015)tolowerthe costofandreduceunnecessaryantimicrobialuse.

Lastly,itwasaimedtodeterminewhetherpharmacydispensing dataaccuratelyreflectantimicrobialconsumption onpaediatric wards,andwhetherdispensingdatacouldbeusedasa‘proxy’for antimicrobialaudits.Comparingthedispensedandtheauditdata, asubstantialvariationbetweenthevolumesofselected intrave-nousantimicrobialsdispensedandtheactualvolumesgivenwas observed.Ingeneral,thedispensingdatarecordedsimilarvolumes ofantimicrobialsusedforthetreatmentofCAIontheward,e.g., ampicillin, amoxicillin–clavulanic acid, and third-generation cephalosporins.However, thepharmacydispensingdata under-estimatedantimicrobialuseforrestrictedantimicrobialsusedfor the treatment of HAI on the ward (i.e., more consumed than dispensedformeropenem,ertapenem,andvancomycin).Themost likely explanation for this finding is that many wards keep commonlyusedantimicrobialsaswardstock;thereforepharmacy orders do not necessarily reflect real-time use. In addition, interviewsofnursingstaffwhoadministerantimicrobialsonthe wardsconfirmedthatsharingofmulti-dosevialsbetweenpatients wascommonpracticeandthatpatientstransferredbetweenwards maybetransferredwiththeirdispensedantimicrobials,obviating theneed toorder stock fromthe pharmacy. In addition,some patientsreceivedtheirfirstdosesofantimicrobialsintheacute

admissionwardpriortotransfertothegeneralpaediatricwardor PICU,whichmayhavealsohadaminoreffectonestimatingtotal antimicrobialvolumeconsumed.

Conversely, the pharmacy dispensed greater volumes of antimicrobialstothePICUthanweregivenaccordingtotheaudit data,exceptforampicillinandthethird-generation cephalospor-ins,whichcorrelatedwellwithobserveduse.Itwassubsequently discoveredthatthepharmacydispensingdatacombinedorders fromboththePICUandtheadjacentneonatalintensiveand high-careunits(NICU),whichislikelythemainreasonforthehigher dispensedvolumes.

Anotherfindingwhenreviewingthepharmacydispensingdata was that broad-spectrum, restricted antimicrobials were com-monly issued as ‘ward stock’, making it impossible to link antimicrobialusewithaparticularpatientandindication.Inview of thesefindings, itis clearthat thepharmacy dispensingdata capturing system would require refinements to improve its usefulnessasatool fortrackingantimicrobialconsumptionand ward stock on the paediatric wards and PICU at the study institution. It should be noted that the pharmacy electronic medicine management system was originally implemented to manage the procurement and finances rather than to track antimicrobialconsumption.Intheinterim,repeatedpoint preva-lencesurveysandantimicrobialusageauditsshouldbecontinued in order to track the impact of antimicrobial stewardship programmes on the wards. However, with some local system adaptations,pharmacy dispensingdatacouldrepresentaviable method for antimicrobial consumption tracking in the future. Electronic prescribing, however, would be ideal to accurately assessantimicrobialusage.

Thestrengthsofthisstudyaretheinclusionofalargeinpatient population from two diverse wards at a large South African children’shospital.Limitationsofthestudyaretheexclusionof patientswhowereadmittedforlessthan48h,whichmayhaveled toeitherover-orunderestimationoftheantimicrobialuse,andthe lack of data on discharge prescriptions (which led to an

Table2

TotalcostsofantimicrobialsutilizedinthepaediatricwardandICU.

Generalpaediatricward PaediatricICU Antimicrobial Grams

used

CostinZARper gram Costin ZAR Costin USD %oftotal cost Antimicrobial Grams used

CostinZARper milligram Costin ZAR Costin USD %oftotal cost Ertapenem 79 368 29095 2447 43% Meropenem 470 132 62114 5224 30% Ampicillin 560 14 8056 678 12% Ertapenem 141 368 51930 4367 25% Amoxicillin 3750 1.9 7125 599 11% Ganciclovir 52 791 40715 3424 19% Aciclovir 25(IV) 60(PO) 235(IV) 1.6(PO) 5853 492 9% Aciclovir 100(IV) 15(PO) 235(IV) 1.6(PO) 23542 1978 11% Meropenem 30 132 4008 337 6% Ampicillin 506 14 7286 613 3% Others – – 13401 1127 20% Others – – 24913 2095 12% Total – – 67541 5680 100% Total – – 210484 17702 100% ICU,intensivecareunit;ZAR,SouthAfricanRand;USD,UnitedStatesDollars;IV,intravenous;PO,oral.

Table3

Variancebetweenselectedantimicrobials:pharmacydispensedversusward/PICUconsumed. Antimicrobial Warddispensed

(grams) Wardconsumed (grams) Variancea PICUdispensed (grams) PICUconsumed (grams) Variancea Meropenem 30 67 0.4 470 289 1.6 Ertapenem 79 48 1.6 141 66 2.1 Vancomycin 3 10 0.3 170 77 2.2 Cefotaxime 262 284 0.9 180 152 1.2 Ceftriaxone 270 345 0.8 234 252 0.9 Ampicillin 560 739 0.8 506 575 0.9 Amoxicillin–clavulanicacid 24 22 1.1 174 62 2.8 PICU,paediatricintensivecareunit.

a

underestimationofthetotaldaysofantimicrobialuse,aschildren couldtakeantimicrobialshome orfinish thecoursein another hospital).Althoughthestudyfindingsarenotgeneralizabletoall hospitalsin Africa,thedatacaninformotherpaediatric centres that are developing antimicrobial consumption surveillance programmes.

Inconclusion,antimicrobialusageratesatTygerbergHospital werevery high compared to developed countryestimates, but comparable to data from the African continent. Antimicrobial consumption as measured by the pharmacy dispensing data, differedfromtheantimicrobialuseobservedinprescriptionaudits, butwithsomesystemadaptations,representsafeasiblemethod forantimicrobialconsumptiontrackinginthefuture.

Acknowledgements

The authors thank Melissa Erasmus (Tygerberg Hospital antimicrobialstewardshippharmacist),TheresaBlockman (Infor-mation Management, Department of Health, Western Cape Government), and the staff and patients of Tygerberg Hospital wardG7andA9PICUfortheirassistance.

Funding

Funding fromthe South African Medical Research Council’s ClinicianResearcherProgrammeandtheDiscoveryFoundation’s Academic Fellowship Award, supported the collection of the antimicrobialprescriptiondata.

Conflictofinterest None.

Authorcontributions

AD and LK developed thestudydesign, performedthedata collectionandanalysis,andproducedthefirstdraft.HF,ED,and AWgave inputonthedata analysisandcriticallyreviewed the manuscript.Allauthorsreadandapprovedthefinalmanuscript. References

AlemnewG,AtnafieSA.Assessmentofthepatternofantibioticsuseinpediatrics wardofDessiereferralhospital,NorthEastEthiopia.IntJMedMedSci2015;7 (1):1–7,doi:http://dx.doi.org/10.5897/ijmms2014.1101.

AmadeoB,ZarbP,MullerA,DrapierN,VankerckhovenV,RoguesAM,etal.European SurveillanceofAntibioticConsumption(ESAC)pointprevalencesurvey2008: paediatricantimicrobialprescribingin32hospitalsof21Europeancountries.J Antimicrob Chemother 2010;65(10):2247–52, doi:http://dx.doi.org/10.1093/ jac/dkq309.

AraujodaSilvaAR,AlbernazdeAlmeidaDiasDC,MarquesAF,BiscaiadiBiaseC, MurniIK,DramowskiA,etal.Roleofantimicrobialstewardshipprogrammesin children:asystematicreview.JHospInfect2017;1:2,doi:http://dx.doi.org/ 10.1016/j.jhin.2017.08.003pii:S0195-6701(17)30447-4.

BlinovaE,LauE,BitnunA,CoxP,SchwartzS,AtenafuE,etal.Pointprevalencesurvey ofantimicrobialutilizationinthecardiacandpediatriccriticalcareunit.Pediatr Crit Care Med 2013;14(6):e280–288, doi:http://dx.doi.org/10.1097/ PCC.0b013e31828a846d.

DeLucaM,DonaD,MontagnaniC,LoVecchioA,RomanengoM,TagliabueC,etal. Antibiotic Prescriptionsand ProphylaxsisinItalian Children.Isittime to change?DatafromtheARPECproject.PLoSOne 2016;11(5)e0154662,doi: http://dx.doi.org/10.1371/journal.pone.0154662.

DillonDG,PirieF,RiceS,PomillaC,SandhuMS,MotalaAA,etal.AfricanPartnership forChronicDiseaseResearch(APCDR).Open-sourceelectronicdatacapture systemofferedincreasedaccuracyandcost-effectivenesscomparedwithpaper methodsinAfrica.JClinEpidemiol2014;67(12):1358–63,doi:http://dx.doi.org/ 10.1016/j.jclinepi.2014.06.012.

DramowskiA,WhitelawA,CottonMF.Burden,spectrum,andimpactof healthcare-associatedinfectionataSouthAfricanchildren’shospital.JHospInfect2016;94 (4):364–72,doi:http://dx.doi.org/10.1016/j.jhin.2016.08.022.

FelekeM,YenetW,LenjisaJL.Prescribingpatternofantibioticsinpediatricwardsof BishoftuHospital,EastEthiopia.IntJBasicClinPharmacol2013;2(6):718–22.

FisherBT,MeaneyPA,ShahSS,IrwinSA,GradyCA,KurupS,etal.Shortreport: AntibioticuseinpediatricpatientsadmittedtoareferralhospitalinBotswana. AmJTropMedHyg2009;81(1):129–31.

GandraS,SinghKS,DasarathaJR,RavishankarK,AshokKC,AnitaS,etal.Point prevalencesurveysofantimicrobialuseamonghospitalizedchildreninsix hospitalsinIndiain2016.Antibiotics(Basel)2017;6(3):19,doi:http://dx.doi. org/10.3390/antibiotics6030019.

GrohskopfLA,HuskinsWC,Sinkowitz-CochranRL,LevineGL,GoldmannDA,Jarvis WR,etal.UseofantimicrobialagentsinUnitedStatesneonatalandpediatric intensivecarepatients.PediatrInfectDisJ2005;24(9):766–73,doi:http://dx. doi.org/10.1097/01.inf.0000178064.55193.1c.

KreitmeyrK,vonBothU,PecarA,BordeJP,MikolajczykR,HuebnerJ.Pediatric antibioticstewardship: successfulinterventions to reduce broad-spectrum antibioticuseongeneralpediatricwards.Infection2017;45(4):493–504,doi: http://dx.doi.org/10.1007/s15010-017-1009-0.

LabiAK,Obeng-NkrumahN,NarteyET,BjerrumS,Adu-AryeeAN,iOfori-AdjeiYA, etal.AntibioticuseinatertiaryhealthcarefacilityinGhana:apointprevalence survey.Labietal.AntimicrobResistInfectControl2018;7:15,doi:http://dx.doi. org/10.1186/s13756-018-0299-z.

MonteiroLGS,ChauqueA,BarrosMP,IraTR.Determinantsofantibioticprescription inpaediatricpatients:thecaseofhospitalsMaputo,Mozambique.SAfrJChild Health 2017;11(3):109–11, doi:http://dx.doi.org/10.7196/SAJCH.2017. v11i3.1224.

National Healthcare Safety Network. Surveillance definition of healthcare-associatedinfectionandcriteriaforspecifictypesofinfectionsintheacute caresetting.January2013(v.17-1).2013Availableat:https://www.cdph.ca.gov/ programs/hai/Documents/Slide-Set-20-Infection-Definitions-NHSN-2013.pdf. [Lastaccessed27February2018].

NewlandJG,BanerjeeR,GerberJS,HershAL,SteinkeL,WeissmanSJ.Antimicrobial stewardshipinpediatriccare:strategiesandfuturedirections. Pharmacother-apy2012;32(8):735–43,doi:http://dx.doi.org/10.1002/j.1875-9114.2012.01155. x.

RattanaumpawanP,BoonyasiriA,VongS,ThamlikitkulV.Systematicreviewof electronicsurveillanceofinfectiousdiseaseswithemphasisonantimicrobial resistance surveillance in resource-limited settings. Am J Infect Control 2017;10:, doi:http://dx.doi.org/10.1016/j.ajic.2017.08.006pii:S0196-6553(17) 30953-7.

Reddy SC, Jacob JT, Varkey JB, Gaynes RP. Antibiotic use in US hospitals: quantification,qualitymeasuresandstewardship.ExpertRevAntiInfectTher 2015;13(7):843–54,doi:http://dx.doi.org/10.1586/14787210.2015.1040766.

SmithMJ,GerberJS,HershAL.Inpatientantimicrobialstewardshipinpediatrics:a systematicreview.JPediatricInfectDisSoc2015;4(4):e127–35,doi:http://dx. doi.org/10.1093/jpids/piu141.

TadesseBT,ElizabethA,OngarelloS,HavumakiJ,WijegoonewardenaM,GonzálezIJ, etal.AntimicrobialresistanceinAfrica:asystematicreview.Tadesseetal.BMC InfectDis2017;17:616,doi:http://dx.doi.org/10.1186/s12879-017-2713-1.

VersportenA,SharlandM,BielickiJ,DrapierN,VankerckhovenV,GoossensH,etal. The antibioticresistance andprescribing inEuropean Childrenproject: a neonatalandpediatricantimicrobialweb-basedpointprevalencesurveyin73 hospitalsworldwide.PediatrInfectDisJ2013;32(6):e242–53,doi:http://dx.doi. org/10.1097/INF.0b013e318286c612.

WorldHealthOrganization.WHO.WHOCollaboratingCentreforDrugStatistics: definitionsandgeneralconsiderations.Geneva:WorldHealthOrganization; 2018Availablefrom: https://www.whocc.no/ddd/definition_and_general_con-sidera/.[Lastaccessed8March2018].