The Value of Long-Lasting Insecticide Nets

12 Pages·2017·1.95 MB·English

International Journal of
Environmental Research
and Public Health
Article
A Meta-Regression Analysis of the Effectiveness of
Mosquito Nets for Malaria Control: The Value of
Long-Lasting Insecticide Nets
Gi-geunYang1,DohyeongKim2,* ID,AnhPham3andChristopherJohnPaul4 ID
1 DepartmentofFireServiceAdministration,WonkwangUniversity,460Iksan-daero,Iksan,
Jeonbuk570-749,Korea;[email protected]
2 SchoolofEconomic,PoliticalandPolicySciences,TheUniversityofTexasatDallas,800WCampbellRoad,
Richardson,TX75080,USA
3 DepartmentofEconomicsandInternationalBusiness,ForeignTradeUniversity,91ChuaLang,DongDa,
Hanoi100000,Vietnam;[email protected]
4 DepartmentofPublicAdministration,NorthCarolinaCentralUniversity,1801FayettevilleSt.,
Durham,NC27707,USA;[email protected]
* Correspondence:[email protected];Tel.:[phone]
Received:12March2018;Accepted:16March2018;Published:19March2018
Abstract: Long-lastinginsecticidalnets(LLINs)havebeenwidelyusedasaneffectivealternative
toconventionalinsecticide-treatednets(ITNs)foroveradecade. Duetothegrowingnumberof
fieldtrialsandinterventionsreportingtheeffectivenessofLLINsincontrollingmalaria,thereisa
needtosystematicallyreviewtheliteratureonLLINsandITNstoexaminetherelativeeffectiveness
andcharacteristicsofbothinsecticidenettings. Asystematicreviewofover2000scholarlyarticles
publishedsincetheyear2000wasconducted. Theoddsratios(ORs)ofinsecticidalneteffectiveness
inreducingmalariawererecorded. Thefinaldatasetincluded26articlesformeta-regressionanalysis,
withasamplesizeof154subgroupobservations. Whilethereissubstantialheterogeneityinstudy
characteristicsandeffectsize,wefoundthattheoverallORforreducingmalariabyLLINusewas
0.44(95%CI=0.41–0.48,p<0.01)indicatingariskreductionof56%,whileITNswereslightlyless
effectivewithanORof0.59(95%CI=0.57–0.61,p<0.01). Ameta-regressionmodelconfirmsthat
LLINsaresignificantlymoreeffectivethanITNsinthepreventionofmalaria,whencontrollingfor
othercovariates. Forbothtypesofnets,protectiveefficacywasgreaterinhightransmissionareas
whennetswereusedforanextendedperiod. However,cross-sectionalstudiesmayoverestimatethe
effectofthenets. Theresultssurprisinglysuggestthatnetsarelesseffectiveinprotectingchildren
undertheageoffive, whichmaybeduetodifferencesinchildbehaviororinadequatecoverage.
Comparedtoapreviousmeta-analysis,insecticide-treatednetsappeartohaveimprovedtheirefficacy
despitetherisksofinsecticideresistance. Thesefindingshavepracticalimplicationsforpolicymakers
seekingeffectivemalariacontrolstrategies.
Keywords: malaria;mosquitonets;long-lastinginsecticidalnets(LLINs);meta-regression
1. Introduction
Withapproximately200millioncasesperyear,malariaisresponsibleforalargeproportionofthe
globalburdenofdiseaseandisaseriouspublichealthissueinAfrica[1]. Thefightagainstmalaria
has forged an extensive global public health initiative. Policymakers consider insecticide-treated
nets(ITNs)oneofthemostsuccessfulvectorcontroltoolstocombatmalaria. Mosquitonetstreated
withsyntheticpyrethroidinsecticideshavebeenshowntocauseadeclineinmalariamorbidityand
mortalityinfieldefficacytrialscarriedoutindifferentcountries[2–5]. Over427millionnetswere
deliveredtohouseholdsinSub-SaharanAfricabetween2012and2014[1].
Int.J.Environ.Res.PublicHealth2018,15,546;doi:10.3390/ijerph15030546 www.mdpi.com/journal/ijerph
Int.J.Environ.Res.PublicHealth2018,15,546 2of12
While the effectiveness of conventional ITNs in controlling malaria has been well
established[4,6,7],therehavebeensomeoperationalproblemshinderingtheirlong-termusefulness.
The effective dose of insecticide on the ITN only lasts for a limited time. Thus, nets need to be
re-treatedregularly(atleastonceperyear)[8]. However,regularre-treatmentatthehouseholdor
communitylevelhasprovenverydifficulttoimplement,resultinginlowre-treatmentrates. Moreover,
repeatedlywashingnetscanreducestheirprotectiveefficacy[9]. InAfrica,itissuggestedthatless
than5%ofdistributedITNswereproperlytreatedorre-treated[10]. Along-termstudyinThailand
reportedthatlowre-treatmentrateswerealimitingfactorinproducingsuccessfuloutcomesforanITN
intervention[11]. DespitenotableeffortsbymanyAfricancountriesandorganizations(suchasthe
GlobalFundandthePresident’sMalariaInitiative)toreducetheincidenceofmalariabyincreasingITN
coveragethroughmassdistribution,thelong-termeffectivenessofITNsformalariacontrolremains
aconcern.
AsasimplersolutiontotheproblemsofconventionalITNs,long-lastinginsecticide-treatednets
(LLINs)havebeendevelopedusingnovelfabrictechnologies[12]. LLINs,recommendedformalaria
controlbytheWorldHealthOrganization,haveinsecticideseitherincorporatedintoorcoatedaround
theirfibers.Thisresistsmultiplewashesandremainseffectiveforalongertime[13].Theselong-lasting
netsaredesignedtomaintaintheirbiologicalefficacyagainstvectormosquitoesforatleastthreeyears
whenusedasrecommended,eliminatingtheneedforregularinsecticidere-treatment. Intermsofthe
costpertreatedyearofprotection,LLINsaresignificantlycheapertousethanITNsorindoorresidual
spraying(IRS).TheannualcostforLLINsrangesfromUSD1.48to2.64,ascomparedtoIRS(USD3.27
to3.90peryear)andstandardITNs(USD1.43to6.05peryear)[14–16].Severaltrialsalsodemonstrated
thesuperiordurabilityofLLINsoverITNs[17,18]. ComparedwithconventionalITNsthataretreated
bybeingdippedintoinsecticide,LLINshaveseveralimportantadvantages. Thisincludesthefactthat
LLINsdonotneedtoundergore-treatment,theirreducedinsecticideconsumptionandtheirreduced
insecticidereleasetotheenvironment.
Manyfield-basedstudiesexaminingtheefficacyofmosquitonetsinreducingtheincidenceof
malarialookatLLINs. However,mostanalysescombinedallinsecticide-treatednets,includingboth
ITNs and LLINs. Some studies examined LLIN efficacy alone, but it is difficult to get a sense
of large-scale performance of LLINs from individual studies. No study has assessed the overall
effectivenessofLLINsinpreventingmalariainfectionratescomparedtoITNs. Anearlymeta-analysis
ofITNefficacy,publishedin1995byChoietal.,reportedthatITNsreducedmalariacasesbyaround
24%[6]. Inhis2004Cochranereviewofinsecticide-treatedbednets,Lengelerfoundarelativerateof
malariaof0.83forchildrenusingITNscomparedtochildrenwhousednonets[4]. Inmanysettings,
LLINs ITNs are used in conjunction with IRS. There is only one published meta-analysis that has
assessed the overall impact of IRS in reducing malaria cases [19]. No meta-regression study has
examinedanypotentialcombinedefficacyofusingbothIRSandLLINs. Fullman(2013)conducted
thefirstmulti-countryanalysisofthecombinedeffectivenessofIRSandinsecticidalnetting,butonly
assessed children under the age of five, and the analysis combined both conventional ITNs and
LLINs[20].
Meta-regressionisaformofmeta-analysisthathasgrowninpopularityinrecentyears. Thisis
in part due to the increasing numbers of systematic reviews and meta-analyses published in the
biomedical literature [21]. Meta-regression is a statistical method of pooling covariate data from
studiesidentifiedinasystematicquantitativeanalysis[22]. Thismethodcalculatesthevariationin
resultsacrossstudiesandidentifiestheextentofvariationcausedbyfactorssuchastargetedgroups,
intervention types, study designs and measurement outcomes. This allows for the assessment of
heterogeneity. Weselectedmeta-regressionasanappropriateapproachtoevaluatetheeffectivenessof
LLINsandITNsinmalariapreventionunderdiverseconditions. Inthispaper,studiespublishedsince
2000onITNsorLLINsweresystematicallyreviewedanddatafromsuitablestudieswereconsolidated
usingmeta-regressionanalysis.Theresultsareusefulformalariapolicymakersinassessingtheefficacy
ofLLINsandITNs,especiallygiventhelimitedresourcesavailabletocombatmalaria.
Int.J.Environ.Res.PublicHealth2018,15,546 3of12
2. Methods
2.1. DatabaseSearchandSystematicReview
Using five databases (PubMed, Google Scholar, Science Direct, JSTOR, and Malaria Journal)
a comprehensive review of the literature published since 2000 was performed. Publications were
searchedforwithkeywordsincludingmalaria,long-lastingtreatedbednet,insecticide-treatedbed
net,LLIN,ITN,prevalenceandincidence. Thisprocessidentified2051articles. Experimentalstudies
thatreportedontheprotectiveefficacyofeitherLLINsorITNsinhumanpopulationsthatadopted
cross-sectional,cohort, orrandomizedcontrolledtrialstudydesignswereincluded. Articleswere
excluded if they met any of the following three criteria: they were review articles; they were
articlesreportingonlytheimpactonmosquitopopulations;theywerearticlesreportingITN/LLIN
interventionscombinedwithuntreatednets,IRS,orimmunization.Afterthisstep,atotalof215articles
wereidentifiedforfullreview. Thefulltextsofthe215articleswerereadtoconfirmtheyqualified
forinclusioninthemeta-analysis. Articleswereexcludedthatdidnotincludesufficientnumerical
information for calculating the effectiveness of ITNs/LLINs, such as the numerical incidence of
malariabeforeandafterintervention(forcohortstudies)orbetweentreatmentandcontrolgroups(for
cross-sectionalstudiesandrandomizedcontrolledtrials). Afterthisstep,atotalof39articlesqualified
fordatacollectionandanalysis.
2.2. DataCollectionandAnalysis
Datafromthe39articleswererecordedwithquantitativemeasuresforthefollowingcovariates:
endemicity,chemicalclass,malariaspecies,interventionduration,groupage,nettypes,studydesign
andstudylocation. Thecostsoftreatednetswerenotrecordedbecausealmostallnetsweresupplied
bysocialdistributioncampaigns,usuallyatasubsidizedprice.Thisapproachpromotesanddistributes
thesociallybeneficialinterventionratherthancommercializingtheproduct. Thirteenarticleswere
excludedduetoalackofquantitativedataforatleastoneofthecovariateslistedabove. Thefinal
sampleformeta-analysisincludes54datapointsfrom12LLIN(long-lastinginsecticide-treatednets)
articlesand100datapointsfrom14ITN(insecticide-treatednets)articles. Figure1depictstheprocess
usedtosearchandselectstudies.
Oddsratios(ORs)wereestimatedfortheeffectivenessoftreatednets. IftheORwasnotreported
in a study, it was calculated based on malaria incidence before and after the intervention (cohort
studies),ormalariaincidencebetweenthetreatmentandcontrolgroups(cross-sectionalstudiesand
randomizedcontrolledtrials). Newvariableswerecreatedfromquantitativeinformationsynthesized
fromstudiestoenableamorecomprehensivemeta-regressionanalysis. Covariatesforsamplesize,
pesticidetype,malariaparasite,location,studyduration,studydesign,andparticipantsundertheage
offivewereincorporatedintothemodel. Samplesizewascreatedbysummingthetotalnumberof
individualsinstudysub-groups. Twodummyvariableswerecreatedfortheinsecticidesdeltamethrin
andalphapermethrin. Thesewerecomparedwiththeinsecticidepermethrinasabaseline. Adummy
variablewascreatedforthetypeofmalariaparasite,where1representsPlasmodiumfalciparumonly,
and0representsbothP.falciparumandPlasmodiumvivax. Adummyvariableforthestudylocationwas
createdtoassesstheeffectivenessoftreatednetsinAfrica,thecontinentbearingthehighestburden
ofmalaria. TorepresentthestudiesundertakeninAfrica,1wasused,and0representedthestudies
undertakeninAsia.Duetoaconcernoverthelong-termdurabilityoftreatednetsandthefactthatITNs
needtoberetreatedatleastonceayeartoensurecontinuedinsecticidaleffectiveness,adummyvariable
wascreatedforthedurationofthestudy,where1representedastudydurationlongerthan12months
and0representedastudydurationoflessthanoneyear. Additionally,twodummyvariableswere
createdtoindicatecohortandcross-sectionalstudydesigns,comparedwithrandomizedcontrolled
trials as baseline. A dummy variable was created for children under the age of five to investigate
themarginalimpactoftreatednetsonthisvulnerablegroup[1,23]. Therewereinsufficientstudies
tocreateadummyvariableforpregnantwomen(anothervulnerablepopulation). Anendemicity
Int.J.Environ.Res.PublicHealth2018,15,546 4of12
covariate was not necessary because almost all studied areas were reported as hyperendemic for
maInlat.r Ji. aE,nveixrocne. pRets.o PnuebliLc HLeIaNlth [220148], 1a5n, dx twoITN studies [25,26] that didnot report endemicity. 4F oinf 1a2l ly,
andmostimportantly,nettypeisadummyvariablewhere1representsLLINonlyand0represents
studies [25,26] that did not report endemicity. Finally, and most importantly, net type is a dummy
ITNonly.
variable where 1 represents LLIN only and 0 represents ITN only.

FiFguigruer1e. M1. eMtheotdhooldooglyoogfys oeaf rscehairncghfionrga fnodr asenldec stienlegcsttiundgi estsutdoiiensc ltuod iencinlumdeet ian-r megerteas-srieognraensasiloynsi s.
analysis.
To conduct a meta-analysis of the odds of malaria prevalence, we implemented a regression
To conduct a meta-analysis of the odds of malaria prevalence, we implemented a regression
meta-analysisusingthestatisticalpackageStata(StataCorp. 2017. StataStatisticalSoftware: Release15.
meta-analysis using the statistical package Stata (StataCorp. 2017. Stata Statistical Software: Release
CollegeStation,TX:StataCorpLLC).TheadjustedR-Squarevaluefortheregressioniscalculatedin
15. College Station, TX: StataCorp LLC). The adjusted R-Square value for the regression is calculated
thestatisticalsoftwarefollowingtheformulaoftheunadjustedR-Squaremultipliedbyaratioofthe
in the statistical software following the formula of the unadjusted R-Square multiplied by a ratio of
degreesoffreedomtotheresidualdegreesoffreedom.
the degrees of freedom to the residual degrees of freedom.
3. Results
3. Results
3.1. DescriptionofArticlesIncludedintheFinalData
3.1 Description of Articles Included in the Final Data
Ofthe14ITNarticles,sixwererandomizedcontrolledtrials[3,5,7,27–29],sixwerecross-sectional
Of the 14 ITN articles, six were randomized controlled trials [3,5,7,27–29], six were cross-
studies[2,25,26,30–32]thatcomparedmalariaprevalencebetweenagroupthatsleptunderITNsand
sectional studies [2,25,26,30–32] that compared malaria prevalence between a group that slept under
agroupthatdidnotandtwowerecohortstudies[33,34]thatanalyzedmalariarateswithingroups
ITNs and a group that did not and two were cohort studies [33,34] that analyzed malaria rates within
groups before and after using an ITN. Twelve articles addressed malaria caused by P. falciparum or
Int.J.Environ.Res.PublicHealth2018,15,546 5of12
beforeandafterusinganITN.TwelvearticlesaddressedmalariacausedbyP.falciparumorP.vivax,
andtwoarticlesdidnotnotethespecies[2,3]. Ofthe12ITNarticlesidentifyingthespeciesofmalaria,
eightarticlesaddressedP.falciparumonly[25–27,30–34]whiletheremainingfourarticlesreportedfor
bothspecies[5,7,28,29],inwhichP.falciparumaccountedforabout80–90%oftotalinfections. Ofthe
14ITNarticles,threewereconductedwithinAsia[5,28,29](82observationsrepresenting82%ofthe
ITNdata)and11trialsexaminedAfricancountries[2,3,7,25–27,30–34](18observationsrepresenting
18%oftheITNdata). Themeanstudysizewas1500participants. Ofallstudies,13%ofobservations
wereinchildrenunderfiveyearsofage[2,3,7,25–27,30,32,34].Twostudiesfocusedonpregnantwomen
from15to39yearsofage[31,33],whichaccountedfor4%ofthetotaldata.
Ofthe12LLINarticles,fivewererandomizedcontrolledtrials[35–39],fourwerecross-sectional
studies [24,40–42], and three were cohort studies [9,43,44]. Among these 12 articles, seven articles
reportedP.falciparumonly[24,38–43],fourarticlesreportedmalariacausedbybothP.falciparumand
P.vivax[9,35,36,44],andonearticledidnotmentionthevectorspecies[37]. Sevenstudiesreporteda
protectiveefficacyforLLINsinAsia[9,24,35–37,39,44](33observationsrepresenting61%oftheLLIN
data)andfivestudieshadstudylocationsinAfricancountries[38,40–43](21observationsrepresenting
39%oftheLLINdata).Themeanstudysizewas1728participants.Atotalof37%ofLLINobservations
lookedatchildrenunderfiveyearsofage[40–43]. Noneofthe12studiesaddressedapopulationof
pregnantwomen. AsummaryofITNandLLINarticlecovariatesisshowninTable1.
Table1.Studydescriptions(countsofarticlecharacteristics).
Covariates ITN LLIN
TotalArticles 14 12
RandomizedControlTrials 6 5
Studytypes Cross-sectional 6 4
Cohort 2 3
Africa 11 5
Studylocations
Asia 3 7
P.falciparumonly 8 7
Malariaspecies
P.falciparumandP.vivax 4 4
Permethrin 6 5
Chemicalclass Deltamethrin 5 5
Alphapermethrin 1 2
Age Childrenunder5 9 4
3.2. Meta-RegressionAnalysis
Theeffectivenessoftreatednetsinreducingtheprevalenceofmalariaisrecordedasanoddsratio
(OR),representingtherelativeoddsofcontractingmalariabetweenatreatmentgroupandacontrol
group. Studiesweresubdividedintotwogroupsforanalysisbasedonthetypeofnet,eitherITNor
LLIN.Figure2depictstheORbetweentheLLINtreatmentandwithout(orbefore)anLLINtreatment,
while Figure 3 depicts a comparison of the OR for the contraction of malaria between ITN users
andITNnon-users. TheoverallsummaryORforLLINeffectivenessinreducingtheprevalenceof
malariais0.44. ThisstatisticallysignificantfindingindicatesthattheoddsofsomeoneintheLLIN
users group contracting malaria were 56% less than in the LLIN non-users group, with the true
populationeffectlyingbetween52%and59%(p=0.0001). Figure2alsoshowstheITNeffectiveness
summaryORis0.59,indicatingtheoddsofcontractingmalariaintheITNgroupwere41%lessthan
in the group without any ITN intervention, with the true population effect between 39% and 43%
(p=0.0001). Thismeta-regressionanalysisrevealsthattheefficacyofbothITNsandLLINsinreducing
theprevalenceofmalariaisstrongandstatisticallysignificant(p=0.0001). Thisstudyalsosuggests
Int.J.Environ.Res.PublicHealth2018,15,546 6of12
thatLLINsaremoreeffectivethanITNsinreducingtheprevalenceofmalaria(asindicatedinthe
Int. J. Environ. Res. Public Health 2018, 15, x 6 of 12
meta-regressionresultfortheLLINcovariate,showninTable2).
Int. J. Environ. Res. Public Health 2018, 15, x 6 of 12

FiguFrigeu2r.eF iF2g.o uFrreoes rt2e.sp tF looprtleosott f optflh oteht eom fm ethteaeta- rm-ereeggtrare-erssessgiioroenns saaionnnaa llayynssiaisls yosofif st thohef e tohodedd odsd srdarsta irotai otoiofo tofhft eht hreee drreueddctuuiocctntiio oonnf omof fammlaarlaaialra iar ia
prevparelevnaclpeernedcveua edleutnoec ett ohd eutheue ts oue stohef eo Luf LsLeIL NoINfs L.sL.( T(INThhse.e r( rToohwwe- -rlloiikkwee- lssiyykmem sbbyoomll fbfooorlr AfoAhrh mAmahdamid a(id2(0i2 1(02210)1 2i2n))di inincddaiitcceaastt etehss atthth ataht teth tlheo ewloleworwe re r
limiltimofit9 o5lfi% m95iCt% oI fCf 9oI5 rf%othr C etIh sfeto usrt dtuhydeyi ss tiuss odsyolo ilsow wsot htlhoaawtti ttithg agoto eietss g bboeeeyys oobnnedydo ttnhhdee tchchhea acrrth taaarrter aea.ra)e.. )a..).

Figure 3. Forest plot of the meta-regression analysis of the odds ratios of the reduction of malaria
FiguFrigeu3r.ep Fr3eo. vrFaeolsertnescptel opdtluoeot ftootf h thtehee mu mseete aot-af r-IerTegNgrrsee.s sssiioonn aannaallyyssiiss ooff tthhee oodddds srartaiotiso sofo tfhteh reedreudctuioctni oonf mofamlaraiala ria
prevparelevnacleendceu edutoe ttoh ethues uesoef oIfT INTNs.s.
Int.J.Environ.Res.PublicHealth2018,15,546 7of12
Table2.Meta-regressionresults(n=154)ofareductioninmalariaprevalencebyLLIN/ITN.
Covariate(DependentVariable=logOR) Coefficient(95%ConfidenceInterval)
UseofLLIN(only) −0.954(−1.419to−0.49)***
Logofinitialprevalence(malariaproportionbeforeintervention) −0.224(−0.355to−0.092)***
Totalsamplesize(thousands) −0.0066(−0.076to0.062)
Studyduration(morethan12months) −0.451(−0.868to−0.034)**
Useofdeltamethrin 0.228(−0.195to0.653)
Useofalphapermethrin 0.399(−0.263to1.063)
Childrenundertheageoffive 1.313(0.821to1.804)***
Plasmodiumfalciparum(only) −0.274(−0.661to0.112)
Cohortstudydesign −0.336(−0.845to0.171)
Cross-sectionalstudydesign −0.7(−1.301to−0.099)**
Constant −0.980(−1.594to−0.367)***
AdjustedR-square 0.33
Note:**p<0.05,***p<0.01.
Eventhoughameta-regressionanalysisrevealedthatLLINshaveagreaterprotectiveefficacythan
ITNs,thereremainsahighlevelofheterogeneitybetweenindividualstudies. AsshowninFigures2
and3,using154observationsfrombothsub-groups,themeta-regressionmodelidentifiessourcesof
heterogeneitysuchastargetgroupcharacteristics,studyconditions,andstudydesigns. IntheLLIN
sub-group, thelargestreportedeffectwasfoundinthestudybyAhmadi(2012)[44], inwhichthe
ORfortheprotectivebenefitofusingLLINswas0.03,whichrepresenteda97%decreaseinmalaria
prevalenceintheLLINusergroupcomparedtotheLLINnon-usergroup. Meanwhile,theLLINOR
wasapproximately0.65inthestudyconductedbyRehmanetal. (2013)[41]. ITNtreatmentshave
shownasimilarrange: ORsfrom0.12[33]to0.71[29]havebeenreported.
In the meta-regression model, the dependent variable is the log transformation of the OR for
netusageinregardtomalariaprevention. TheprimaryindependentvariablesarewhetheranLLIN
wasused,thelog-transformedproportionofstudypopulationwithmalariabeforetheintervention,
thetotalsamplesize,whetherthestudydurationlastedlongerthan12months,childrenundertheage
offive,theuseofdeltamethrinandalphapermethrin,P.falciparumonly,andcohortandcross-sectional
study designs. An OR of significantly less than 1 or a negative log OR implies that an increase in
a covariate is associated with a decrease in the risk of contracting malaria. Use of LLINs, the log
transformationoftheinitialprevalencepriortointervention,whetherthestudylastedlongerthan
12monthsandwhetherchildrenlessthanfiveyearsofagewerepresentinthestudyarevariablesthat
werehypothesizedtobenegativelyrelatedtotheOR.AmodelusingtheAfricancovariatehadalower
adjustedR-square,aswellasaninsignificantcoefficientforAfrica. Therefore,theAfricancovariate
wasnotincludedinthefinalmodel. Thebest-fitmodelisshowninTable2.
Ashypothesized,LLINusage,initialmalariaprevalence,andthestudydurationwereshownto
bestrongandstatisticallysignificantnegativepredictorsoftheORformalariaprevalence. Overall,
themeta-regressionresultscanbeinterpretedasfollows: Firstly,thenegativerelationshipofinitial
malariaprevalencetothelogtransformationoftheORindicatedthattheuseofLLIN/ITNconfersa
strongerprotectionagainstcontractingmalariaforpeoplelivinginmalaria-endemicareas. Second,
thenegativerelationshipofthelogtransformationoftheORtowhetherthestudylastedlongerthan
12monthsimpliesthatinsecticidalnetswereeffectiveoverlongerperiodsoftimeiftheyareeither
long-lastingorroutinelyre-treated. Asnoted,conventionalITNsneedstobere-treatedatleastoncea
yeartomaintaininsecticidaleffectiveness. Third,contrarytoexpectation,thecoefficientforchildren
under the age of five was positive and significant, indicating that treated nets are less effective in
preventingmalariaforthisagegroupcomparedtochildrenagedfiveyearsorover.However,netswere
somewhateffectiveinreducingmalariaprevalenceinbothagecategories.Fourth,cross-sectionalstudy
designwasnegativelyrelatedtotheOR,ascomparedtocohortstudyandrandomizedcontrolledtrial
studydesigns,whichcouldsuggestthatcross-sectionalstudiesoverestimatedtheeffectofmosquito
nets[45]. Fifth,accordingtothemeta-regressionresults,thedummyvariableofLLINusagehada
Int.J.Environ.Res.PublicHealth2018,15,546 8of12
statisticallysignificanteffectonthelogtransformationoftheORwhencontrollingforothervariables.
ThisindicatesthatLLINusageconfersgreaterprotectionagainstmalarialinfectionthanITNusage.
Finally,theanalysiscontrolledforanyeffectoftotalsamplesize,differentmalariatypes(P.falciparum
andP.vivax),anddifferentinsecticides(deltamethrin,alphapermethrinandpermethrin). Therewas
nodifferenceinnetefficacyforthesecovariates.
4. Discussion
Inthemeta-analysisstudypublishedChoietal.,ITNswereshowntoavertaround24%ofmalaria
cases[6]. Thepresentstudy,usingonlyliteraturesincetheyear2000,providesacomparisonwith
findingsin1995. Thismeta-analysisdemonstrateshowmuchITNefficacyhasimprovedaswellas
examiningthenoveltechnologiesandmaterialsofLLINs. Thismeta-regressionfindsthattheoverall
effectivenessofITNsandLLINsinreducingtheincidenceofmalariais41%and56%respectively,
when compared to people who used no form of netting. This implies that the efficacy of treated
netshasimprovedsubstantiallyoverthelastdecade,despiteconcernsthatpyrethroidresistancein
mosquitoesmaycompromisetheabilityoftreatednetstokillmosquitoes.
TherearesomedifferencesbetweenChoietal.’sstudyandthepresentstudy. Inthe1995study,
eightofthetentotalfieldtrialsincludedintheanalysislimitedthestudypopulationtochildrenless
than15yearsold. TheothertwostudiesincludedinChoietal.’sanalysiscontainedbothchildren
andadults. Choietal.’sanalysisusedtwosub-groupsbasedonthetypeofcontrol. Thesewerean
ITNinterventiongroupversusagroupthatreceivedanuntreatedbednet,andagroupthatreceived
permethrin-impregnatedbednetsascomparedtoagroupthatreceivednonet. Eachsub-groupused
smallsamplesofsixfieldtrials,comparingmeasuresoftheincidenceofparasitemiaasafunctionof
person-weeksatriskforparasitemia. Inthisanalysis,thesamplestudysizeincreasedto26studies
(published since the year 2000), in which 14 studies reported on the effectiveness of ITNs and 12
reportedontheeffectivenessofLLINs. ThisanalysisfocusedontheprotectiveefficacyofbothLLINs
andITNsandtheirprotectiveefficacywhenusedinconjunctionwitheachother. Thisstudyusedthe
malariaprevalenceoddsratioasthemainindicatorofinfectionprevention. Inthisanalysisandunlike
Lengeler’s[4],thetargetpopulationisnotlimitedtochildren,andincludesanalysisfromtheentire
studypopulationatrisk.
Meta-analyses, such as the one presented here, are particularly useful for policymakers in
strategically allocating resources in disease control. The main finding of this meta-analysis is the
outstandingprotectiveefficacyofLLINsoverITNsoverlongperiodsoftime. LLINsaredesignedso
thattheirinsecticidalactivitycouldlastaslongasthenetsthemselves,aclaimwhichissupportedby
thismeta-analysis.Anadditionalfindingisthateventhoughsomestudiesshowthatvulnerablegroups
(infants,youngchildren,andpregnantwomen)sleepunderinsecticidalbednettingmoreoftenthan
others[46,47],themeta-regressionresultsfindthatinsecticidenetsdonotconferahigherprotective
efficacyforchildrenundertheageoffive. Eventhoughinregardtothedistributionofinterventions,
priorityisgiventothemostvulnerable,childrenofvaryingagesoftensleeptogether. Netsarenot
designed to cover multiple children and thus, coverage is often inadequate. Besides, nets may be
uncomfortabletouseinlimitedspacesandhot,humidclimates,whichcanresultinreduceduseof
bednets[48]. Asinfantsandyoungchildrenoftensleepwiththeirmothers,adultsleepingpreferences
canaffectchildren’sexposure. Duringthedryseasoninmanyplaces,itiscommonforpeopletosleep
unprotectedforatleasthalfthenight[49,50]. Additionally,evidencesuggeststhatthepossessionand
appropriateuseoftreatednettingdoesnotautomaticallymeanthatthesenetsareusedcorrectly[51,52].
Thefrequentperceptionofmalariaasacommondiseaseleadspeopletobelievethatitisnotworth
theinconvenienceofdailynetuse. Recipientshaveevenre-solddonatedbednets. Thesereasonsare
consistentwiththedishearteningreportthatin2005inAfrica,itwasestimatedthatonly3%ofchildren
undertheageoffivesleptunderinsecticide-treatednets[26,53]. Arecentstudyreportsthatonly14%
ofITNswereusedbychildrenundertheageoffive[52],whichisconsistentwiththeknownbarriersto
thecorrectuseofITNsmentionedabove.Toincreasemotivationformalariaprevention,populationsin
Int.J.Environ.Res.PublicHealth2018,15,546 9of12
endemicareasneedtobewelleducatedabouttheseverityofmalariaandthevalueofbednettingasan
infectionpreventionstrategy. Thisrequiresbettercommunicationbetweenauthorities,socialscientists,
healthprofessionalsandtargetpopulations. Moreover,moreefficientcommunity-basedstrategiesto
preventmalariainfectionaredesperatelyneededinhighendemicandruralareas[26].
The limitations of this study are common to meta-analyses. A meta-analysis of the literature
only examines published studies. The comprehensive effectiveness of LLINs and ITNs might not
be captured if most published studies focus on success stories, while insignificant impacts are
not published. Several studies were excluded due to missing selection criteria for the adopted
meta-regressionmodel,whichmayhaveledtobias. Also,thestudiesusedinthemeta-analysisvary
byapproach. Forexample,mostreportedthatthedistributionofnettingwasfree,thoughsomestudies
involvedthepromotionand/ormarketingofnets. Thepriceofnetsthatusershavetopaymayaffect
properuseanddeployment[54]. Morebroadly,thereissubstantialheterogeneityintheparameter
measurements of the included studies. It is therefore difficult to determine to which populations
wecanapplyourfindings. Varioussubgroupanalysesbasedonstudydesigns,studyconditionsor
targetgroupcharacteristicscouldfurtherexplorethecriticalsourcesofheterogeneity. Furthermore,
many studies failed to document coverage rates, even though high coverage rates are needed to
realize the full potential of treated nets. In other words, the finding is based on an “intention
to treat” analysis—an analysis of people who could be covered by the distributed nets, not an
analysisofpeoplewhoareconfirmedtobeusingthenets. Therefore,potentialeffectivenesscouldbe
underestimatedduetoanassumptionthatallhouseholdmemberssleptunderthedistributednets.
Finally,themeta-regressionmodelR-squareof33%isnothighandthesamplesizeof154datapoints
isquitesmall. However,afindingbasedonan“intentiontotreat”ascomparedtothetrueratewould
resultinalowerR-square. Smallsamplesizeisacommonchallengeformeta-regressionstudies[6,19].
Moreover,theperformanceoftreatednetscanvarydependingonthesettingandconditions.
Thereremainshighglobaldemandforsustainablemalariacontrolandelimination. Providedthat
peopleusetreatednetsproperlyandconsistently,LLINsandITNsarehighlyeffectiveatpreventing
malaria [1,4]. Replacing ITNs with LLINs and achieving universal access to LLINs would greatly
advancemalariacontrolforallpopulations[55]. ManymillionsofconventionalITNsremaininroutine
use. However,itisimportanttobalancetheequityofmaintainingwidespreadITNusagewhilealso
makingamoreintensifiedefforttoreplaceITNswithLLINs.
5. Conclusions
Themeta-regressionmodelpresentedinthispaperconfirmsthatLLINsarestatisticallymore
effectivethanITNsinpreventingmalaria,althoughtheeffectivenessofbothhasbeensubstantially
improved during the past decade. The findings support the importance of treated nets, and the
improvementfromLLINs,inmalariacontrol.
Acknowledgments:ThispaperwassupportedbyWonkwangUniversityin2017.
AuthorContributions:G.Y.andD.K.designedthestudyandprovidedoperationalguidance;A.P.analyzedand
interpretedthedata;A.P.andD.K.draftedthemanuscript.C.P.criticallyrevisedthemanuscriptforintellectual
content.Allauthorsreadandapprovedthefinalmanuscript.G.Y.andD.K.areguarantorsofthepaper.
ConflictsofInterest:Theauthorsdeclarenoconflictofinterest.
References
1. World Health Organization (WHO). World Malaria Report 2016; World Health Organization: Geneva,
Switzerland,2016.
2. Abdulla,S.;Schellenberg,J.A.;Nathan,R.;Mukasa,O.;Marchant,T.;Smith,T.;Tanner,M.;Lengeler,C.
Impactonmalariamorbidityofaprogrammesupplyinginsecticidetreatednetsinchildrenagedunder
2yearsinTanzania:Communitycrosssectionalstudy.BMJ2001,322,270–273.[CrossRef][PubMed]
Int.J.Environ.Res.PublicHealth2018,15,546 10of12
3. Howard,S.C.;Omumbo,J.;Nevill,C.;Some,E.S.;Donnelly,C.A.;Snow,R.W.Evidenceforamasscommunity
effectofinsecticide-treatedbednetsontheincidenceofmalariaontheKenyancoast. Trans. R.Soc. Trop.
Med.Hyg.2000,94,357–360.[CrossRef]
4. Lengeler,C.Insecticide-treatedbednetsandcurtainsforpreventingmalaria.CochraneDatabaseSyst.Rev.
2004,CD000363.[CrossRef][PubMed]
5. Rowland, M.; Webster, J.; Saleh, P.; Chandramohan, D.; Freeman, T.; Pearcy, B.; Durrani, N.; Rab, A.;
Mohammed,N.PreventionofmalariainAfghanistanthroughsocialmarketingofinsecticide-treatednets:
Evaluationofcoverageandeffectivenessbycross-sectionalsurveysandpassivesurveillance. Trop. Med.
Int.Health2002,7,813–822.[CrossRef][PubMed]
6. Choi, H.W.; Breman, J.G.; Teutsch, S.M.; Liu, S.; Hightower, A.W.; Sexton, J.D. The effectiveness of
insecticide-impregnated bednets in reducing cases of malaria infection: A meta-analysis of published
results.Am.J.Trop.Med.Hyg.1995,52,377–382.[CrossRef][PubMed]
7. Hawley,W.A.;Phillips-Howard,P.A.;terKuile,F.O.;Terlouw,D.J.;Vulule,J.M.;Ombok,M.;Nahlen,B.L.;
Gimnig,J.E.;Kariuki,S.K.;Kolczak,M.S.;etal.Community-wideeffectsofpermethrin-treatedbednetson
childmortalityandmalariamorbidityinwesternKenya.Am.J.Trop.Med.Hyg.2003,68(Suppl.4),121–127.
[PubMed]
8. WorldHealthOrganization(WHO).InstructionsforTreatmentandUseofInsecticide-TreatedMosquitoNets;
WorldHealthOrganization:Geneva,Switzerland,2002.
9. Dev,V.;Phookan,S.;Padhan,K.;Tewari,G.G.;Khound,K.Laboratorywash-resistanceandfieldevaluation
ofdeltamethrinincorporatedlong-lastingpolyethylenenetting(Netprotect(R))againstmalariatransmission
inAssam,north-eastIndia.ActaTrop.2011,119,172–177.[CrossRef][PubMed]
10. Dabire, R.K.; Diabate, A.; Baldet, T.; Pare-Toe, L.; Guiguemde, R.T.; Ouedraogo, J.B.; Skovmand, O.
Personalprotectionoflonglastinginsecticide-treatednetsinareasofAnophelesgambiaes.s.resistanceto
pyrethroids.Malar.J.2006,5,12.[CrossRef][PubMed]
11. Patipong,S.;Yongchaitrakul,S.FieldefficacyandpersistenceofLongLastingInsecticidetreatedmosquito
Nets(LLINs)incomparisonwithconventionalInsecticideTreatedmosquitoNets(ITN)againstmalaria
vectorinThailand.J.VectorBorneDis.2008,5,7–13.
12. Guillet,P.;Alnwick,D.;Cham,M.K.;Neira,M.;Zaim,M.;Heymann,D.;Mukelabai,K.Long-lastingtreated
mosquitonets:Abreakthroughinmalariaprevention.Bull.WorldHealthOrgan.2001,79,998.[PubMed]
13. WorldHealthOrganization(WHO).WHORecommendationsforAchievingUniversalCoveragewithLong-Lasting
InsecticidalNetsinMalariaControl;WorldHealthOrganization:Geneva,Switzerland,2014;Availableonline:
http://www.who.int/malaria/publications/atoz/who_recommendation_coverage_llin/en/(accessedon
23September2017).
14. Yukich, J.; Tediosi, F.; Lengeler, C. Comparative Cost-Effectiveness of ITNs or IRS in Sub-Saharan Africa;
NewsletterJuly2007;SwissTropicalInstitute:Basel,Switzerland,2007.
15. Programme,G.M.Insecticide-TreatedMosquitoNets:AWHOPositionStatement;WorldHealthOrganization:
Geneva,Switzerland,2007.
16. Yukich, J.O.; Lengeler, C.; Tediosi, F.; Brown, N.; Mulligan, J.A.; Chavasse, D.; Stevens, W.; Justino, J.;
Conteh,L.;Maharaj,R.;etal.Costsandconsequencesoflarge-scalevectorcontrolformalaria.Malar.J.2008,
7,258.[CrossRef][PubMed]
17. Graham, K.; Kayedi, M.H.; Maxwell, C.; Kaur, H.; Rehman, H.; Malima, R.; Curtis, C.F.; Lines, J.D.;
Rowland, M.W. Multi-country field trials comparing wash-resistance of PermaNet and conventional
insecticide-treatednetsagainstanophelineandculicinemosquitoes. Med. Vet. Entomol. 2005,19,72–83.
[CrossRef][PubMed]
18. Sreehari,U.;Raghavendra,K.;Rizvi,M.M.;Dash,A.P.Washresistanceandefficacyofthreelong-lasting
insecticidalnetsassessedfrombioassaysonAnophelesculicifaciesandAnophelesstephensi. Trop. Med.
Int.Health2009,14,597–602.[CrossRef][PubMed]
19. Kim, D.; Fedak, K.; Kramer, R. Reduction of malaria prevalence by indoor residual spraying:
Ameta-regressionanalysis.Am.J.Trop.Med.Hyg.2012,87,117–124.[CrossRef][PubMed]
20. Fullman, N.; Burstein, R.; Lim, S.S.; Medlin, C.; Gakidou, E. Nets, spray or both? The effectiveness of
insecticide-treatednetsandindoorresidualsprayinginreducingmalariamorbidityandchildmortalityin
sub-SaharanAfrica.Malar.J.2013,12,62.[CrossRef][PubMed]
21. Haidich,A.B.Meta-analysisinmedicalresearch.Hippokratia2010,14(Suppl.1),29.[PubMed]