Unit 1 Fire Safety
RESEA RCHARTICL E Evaluating implementation ofafire- prevention injuryprevention briefingin children's centres:Clusterrandomised controlled trial Toity Deave 1 , Adrian Hawkins 2 , Arun Kumar 3 , Mike Hayes 4 , Nicola Cooper 5 , Michael Watson 6 , Joanne Ablewhite 3 , Carol Coupland 3 , Alex Sutton 5 , Gosia Majsak- Newman 7 , Lisa McDaid 8 , Trudy Goodenough 1 , Kate Beckett 9 , Elaine McColl 2,10 , Richard Reading 11 , Denise Kendrick 3 1 Centre forChild &Adolesce ntHealth, Health&Applied Sciences, University ofthe West ofEngland Bristol, Bristol, UnitedKingdom ,2 Institute ofHealth &Society, Baddiley-Clar kBuilding, Newcastle University, Newcastle uponTyne, United Kingdom ,3 Division ofPrimary Care,School ofMedicine, Universityof Nottingham ,Nottingham, UnitedKingdom ,4 Child Accident PreventionTrust, Barnet, London,United Kingdom ,5 Department ofHealth Sciences, UniversityofLeicester, Leicester, UnitedKingdom, 6Faculty of Medicine andHealth Sciences, SchoolofHealth Sciences ,University ofNottingham ,Nottingham ,United Kingdom ,7 Norfolk andSuffolk Primary andCommun ityCare Resea rchOffice, Hosted bySouth Norfolk CCG, Norwich ,United Kingdom, 8Norfolk andNorwich University Hospital,NHSClinical Researc hand Trials Unit,Norwich MedicalSchool,University ofEast Anglia, Norwich ,United Kingdom, 9University ofthe West ofEngland, Researchand Innovation ,University HospitalsBristol NHSFounda tionTrust, Educatio n Centre, Bristol,UnitedKingdom ,10 Newcastle ClinicalTrialsUnit,Newcastle University, Newcastleupon Tyne, United Kingdom, 11Jenny LindPaediat ricDepartment, NorfolkandNorwich UniversityHospital, Norwich ,United Kingdom denise.ken [email protected] k Abstract Background Many developed countrieshavehighmortality ratesforfire-related deathsinchildren aged 0±14 years withsteep social gradients. Evidence-based interventionstopromote firesafety practices exist,butthe impact ofimplementing arange ofthese intervention sin children's services hasnotbeen assessed. Wedeveloped anInjury Prevention Briefing(IPB),which brought together evidence abouteffective firesafety interventions andgood practice in delivering interventions; plustraining andfacilitation tosupport itsuse and evaluated its implementation.
Methods We conducted acluster randomised controlledtrial,withintegrated qualitative andcost- effectiveness nestedstudies, acrossfourstudy sitesinEngland involving children's centres in disadvantaged areas;participants werestaffandfamilies attending thosecentres. Cen- tres were stratified bystudy siteand randomised withinstratatoone ofthree arms: IPBplus facilitation (IPB+),IPBonly, usual care.IPB+centres received initialtraining andfacilitation at months 1,3,and 8.Baseline datafrom children's centreswerecollected betweenAugust PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 1/23 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 2 3 ( 1 $ & & ( 6 6 Citation: DeaveT,Hawkins A,Kumar A,Hayes M, Cooper N,Watson M,etal. (2017) Evaluating implementa tionofafire-prev entioninjury preventio nbriefing inchildren's centres:Cluster randomised controlledtrial. PLoS ONE12(3):
e0172584. https://doi.org/10.1371/j ournal.
pone.017258 4 Editor: Belinda JGabbe, Monash University, AUSTRALIA Received: January12,2016 Accepted: February5,2017 Published: March24,2017 Copyright: 2017 Deave etal. This isan open access articledistributed undertheterms ofthe Creative Commons Attribution License,which permits unrestricte duse, distribu tion,and reproduction inany medium, providedtheoriginal author andsource arecredited.
Data Availabilit yStatement: Dataareavailable within thepaper anditsSupport ingInformation files. Participants werenotasked onthe consent form toprovide consent todata sharing, sofurther patient dataisnot publicly available. Formore information, pleasecontact thecorrespon ding author atdenise.kend rick@nottingha m.ac.uk.
Funding: Thispaper presents independent research fundedbythe National InstituteforHealth Research (NIHR)underitsProgramme Grantsfor 2011 andJanuary 2012andfollow-up datawere collected betweenJune2012 andJune 2013. Parent baseline datawere collected betweenJanuary2012andMay 2012 andfollow- up data between May2013 andSeptember 2013.Datacomprised baselineand12month parent- andstaff-complete dquestionnaires, facilitationcontactdata,activity logsandstaff interviews. Theprimary outcome waswhether familieshadaplan forescaping fromahouse fire. Treatment armswerecompared usingmultilevel modelstoaccount forclustering by children's centre.
Results 1112 parents at36 children's centresparticipated. Therewasnosignificant effectofthe intervention onfamilies' possession ofplans forescaping fromahouse fire(adjusted odds ratio (AOR) IPBonly vs.usual care:0.93, 95%CI 0.58,1.49;AORIPB+ vs.usual care1.41, 95%CI 0.91,2.20). However, significantly morefamilies inthe intervention armsreported more behaviours forescaping fromhouse fires(AOR IPBonly vs.usual care:2.56,95%CI 01.38, 4.76;AORIPB+ vs.usual care1.78, 95%CI 1.01,3.15).
Conclusion Our study demonstrated thatchildren's centrescandeliver aninjury prevention intervention to families indisadvantaged communitiesandachieve changes inhome safety behaviours.
Introduction Childhood fire-related deathsandinjuries areanimportant globalissue[1] andareone ofthe leading causesofdeaths forchildren under14years inthe US.[2] In2009, almost 119,000 U.S.
children wereinjured severely enoughduetounintentional firesandburns thatthey hadto visit anED. Fires andburns areone ofthe major causes ofnonfatal unintentional injuriesin children inthe US. [2]Compared withother high-income countries,theUK hashigh mortal- ity rates fordeaths fromfireand flames inchildren aged0±14 years withsteep social gradients in mortality.[1, 3,4] Some interventions areeffective inreducing therisk offire-related injuryandinpromoting fire-prevention practices.[5±8]Smokealarmscanreduce therisk ofdeath inhouse fires.[5, 6] Providing educationandsmoke alarms canincrease theprevalence ofworking smokealarms and educational interventions canincrease theprevalence ofplans forescaping fromhouse fires.[7, 8]Although evidence-based interventionstopromote firesafety practices exist,trans- lating research findingsintopractice doesnotalways occur.Injury prevention programmes are unlikely toachieve theiraimsifthey arenot effectively implemented. Knowinghoweffec- tive interventions canbeimplemented morewidely isamajor challenge ininjury prevention.
To address thistranslational gap,wedeveloped afire prevention intervention comprisingan Injury Prevention Briefing(IPB)[9] thatcombined guidanceonbest practice fordelivering injury prevention programmes inareal-world settingwithevidence ofeffectiveness ofinter- ventions, alongwithtraining andafacilitation packagetosupport implementation ofthe IPB.
The IPBwasdesigned foruse bychildren's servicessuchaschildren's centresinthe UK or Head Startprogrammes inthe US. These services aimtoimprove outcomes foryoung children and toreduce inequalities inhealth, withaparticular focusonthe most disadvantaged.[10±12] They provide community-based services,information andsupport forfamilies. Children's Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 2/23 Applied Researc hProgramme (RP-PG-0407- 10231). Theviews expressed inthis article are those ofthe authors andnotnecessarily thoseof the NHS, theNIHR orthe Departmen tof Health.
The institutions ofthe authors therefore received funds forustocarry outtheresearch. Thefunders had norole instudy design, datacollecti onand analysis, decisiontopublish, orprepara tionofthe manuscript.
Competing interests:The authors havedeclared that nocompeting interests exist. centres focusonthose withpre-school children,similartothose inHead Start,aUS govern- ment-funded programmethatserves lowincome families withchildren between3-5years.[13] The services includehomesafetyinterventions andserve populations atparticular riskoffire- related injury.However, theeffectiveness ofdelivering arange offire prevention activities, such asthose included inthe IPB insuch settings isunknown. Inthis paper, wereport onthe evaluation ofthe effectiveness andcost effectiveness ofimplementing theIPB inchildren's centres inthe UK.
Methods Design andsetting We undertook athree-arm multi-centre clusterrandomised controlledtrial(RCT), withinte- grated economic evaluation, infour sites inEngland (Nottingham, Bristol,Norwich andNew- castle). Weused acluster RCTtoprevent contamination betweenfamilieswhoattended the same children's centreandbecause intervention deliverywasmore pragmatic atchildren's centre level.Embedded qualitativeinterviewswithkeychildren' staffatthese sitesprovided additional contextual information regardingfacilitators andbarriers toIPB implementation.
[14] Fulldetails ofthe methods used[15] andofthe qualitative analysis[14] arereported elsewhere.
Participants Children's centreswereinvited toparticipate iftheir catchment areahadmore than50%of under-5 year-olds livinginone ofthe 30% most disadvantaged superoutput areasinEngland, which aregeographical areasforthe collection andpublication ofsmall areastatistics.[16] Recruitment tookplace between 03/08/2011 and10/01/2012.
Families livinginthe children's centrecatchment area,whohadattended thecentre inthe previous threemonths withparents agedatleast 16years andachild under threeyears, were eligible. Recruitment tookplace between 05/01/2012 and31/05/2012. Therecruitment strate- gies aredescribed elsewhere.[15, 17]For both children's centresandfamilies, recruitment was defined asproviding writtenconsent andcompletion ofbaseline questionnaire. Allpartici- pants completed aconsent form.
Intervention The intervention wasdeveloped usingtheUK Medical Research Council(MRC)guidance for the development andevaluation ofcomplex interventions[18] andincluded identifying the evidence-base, identifyingappropriate theoryandmodelling processesandoutcomes. Evi- dence abouttheeffectiveness ofinterventions wasascertained fromasystematic reviewof interventions toprevent homeinjuries, including thosefromhouse fires[7] andasystematic review offacilitators andbarriers forhome injury prevention interventions forpre-school children.[19] Evidenceaboutthedesign, content anddelivery ofthe intervention camefrom several sources, including theHealth Development Agency`Effective ActionBriefing' forput- ting evidence intopractice forthe promotion ofdomestic smokealarms[20] andareview of reviews ofliterature onthe implementation andfacilitation ofhealth promotion interventions, undertaken aspreliminary workforthis trial. Wealso interviewed nationalandlocal leaders, undertook workshops, oneineach trialsite,with community practitioners, staffinchildren's centres, fromtheFire andRescue Service, NHSandcommissioners. Bydoing this,webrought together thescientific evidenceonwhat works, withbestpractice fromthose whodeliver injury prevention programmes inthe community.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 3/ 23 A number ofcommon themesemerged fromthereview ofreviews; fourreviews hadcon- vergent findings abouttheexplanatory factorsthataffect theimplementation ofcommunity prevention programmes; oneofwhich wastraining andtechnical assistance.[21±24] However national guidance documents, suchasthose fromtheNational InstituteforCare Excellence, are often provided inthe UK without suchsupport. Thetrial therefore comprised twointer- vention arms,onewhere theIPB wasprovided withtraining andfacilitation (describedbelow) the other provided theIPB without training andfacilitation.
The intervention wasdeveloped basedonfive behavioural changetheories (healthbelief model, socialcognitive theory,theoryofreasoned action,theoryofself-regulation andself- control andtheory ofsubjective cultureandinterpersonal relations)fromareview ofbehav- iour change theories forinjury prevention.[25] Ourintervention aimedtoaddress thethree factors described asnecessary andsufficient forproducing abehaviour changebyhelping par- ticipants (bothchildren's centrestaffandfamilies) formintentions tochange behaviour, remove environmental barriersandprovide participants withtheknowledge andskills toper- form thebehaviour.
There weretwointervention armsandoneusual carearm:
1. IPB delivery, athree hourstafftraining sessionandon-going facilitation tosupport imple- mentation (IPB+); 2. IPB posted tochildren's centre(IPBonly); 3. Children's centreundertake usualfireprevention activity(usualcare).
The IPBwasdeveloped specifically foruse inchildren's centres[9] usingaseven-step pro- cess.[20, 26]The IPBprovided advice,information andactivities tosupport delivery offive key firesafety messages: smokealarmuseand maintenance, plansforescaping fromahouse fire, potential causesofhouse fires,safestorage ofmatches andlighters andbedtime firesafety routines. Thetraining provided information onfire-related injuries;development, principles and content ofthe IPB; practice inusing theIPB and inthe development ofthe IPB implemen- tation plan.
The facilitation comprisedtelephoneorface-to-face contactsfromtheresearch teamat1,3 and 8months tocollect information onimplementation progress,addressquestions and barriers toimplementation andprovide advice,examples ofgood practice, aresource listand contacts withother organisations, e.g.,fireand rescue service. Children's centresvaryconsid- erably intheir management andoperational processesandinthe populations theyserve but they areexperienced indelivering healthpromotion programmes. Forthis reason, andto ensure theimplementation ofthe IPB reflected thereal-world setting,theywere asked to develop aplan forimplementing theIPB which wasmost suitable totheir circumstances and those ofthe families theyserve. Ifthey were unable todeliver allfive fire-safety messages,they were asked tofocus onsmoke alarms andfireescape plansasthese havethestrongest evi- dence-base. Children'scentresusedtheir usual centre processes fordisseminating information about theIPB tostaff, bothcurrent andnew.
Outcomes The primary outcome wasafamily levelbinary variable forwhether ornot thefamily hada plan forescaping fromahouse fire.Secondary outcomesaredescribed inBox 1:
Outcomes wereascertained 12months post-intervention commencementinthe IPB+ and IPB only arms and12months post-randomisation inthe usual carearm, plusfacilitation Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 4/ 23 contact datawhich werecollected at1,3 and 8months postcommencement ofthe intervention in the IPB+ arm.
Outcomes weremeasured usingarange oftools. Families completed baselineand12 month follow-up questionnaires. Datafrom children's centresincluded baselineandpost- intervention manager-completed questionnaires,facilitationquestionnaires andinterviews with theIPB+ children's centrestaffat1,3and 8months andactivity logstorecord home safety activities. Thetwointervention armsalsocompleted questionnaires andinterviews to assess implementation fidelityat12 months.[14] Baselinedatafrom children's centreswere Box 1.Secondary outcomes Family outcomes 1. Intervention familyreported morefireescape behaviours thanthose families inthe control group(using abinary measure derivedfromfivecomponent itemsusing latent variable analysis. Thecomponent itemsare:having doorkeysaccessible, hav- ing window lockkeys accessible, havingatorch beside thebed, knowing thesound of asmoke alarmandhaving exitsclear); 2. Family hadsmoke alarms fittedandworking onevery leveloftheir home; 3. Family reported fire-setting ormatch playbytheir children; 4. Family reported bedtimefiresafety routine score; 5. Family hadaccessed smoking cessation services; 6. The number ofcorrect responses tofire safety knowledge questions; 7. Family reported beingfairlysatisfied orvery satisfied withhome safetyinformation provided bychildren's centres; 8. Implementation ofthe IPB assessed by:
a. Family hadreceived adviceoneach ofthe 5key messages inthe IPB inthe last year; b. Family hadattended afire safety session inthe last year; c. The number offire safety sessions attended byfamily inthe last year; d. Family hadattended afire safety session atachildren's centreinthe last year; e. Family hadattended sessionsabouteachofthe 5key messages inthe IPB inthe last year; 9. Family's resource-use andexpenditure inrelation tofire safety practices.
Children's centreoutcomes 10. Children's centreprovided information andadvice onfire prevention; 11. Resource useand expenditure incurredinrelation tofire prevention activities 12. Reported implementation ofthe IPB within children's centres; 13. Barriers andfacilitators tochildren's centresimplementing theIPB.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 5/ 23 collected betweenAugust2011andJanuary 2012andfollow-up data,including facilitation contact datawere collected betweenJune2012 andJune 2013. Parent baseline datawere col- lected between January2012andMay 2012 andfollow-up databetween May2013 andSep- tember 2013.Follow-up datawere analysed in2014.
Randomisation Children's centreswerestratified bystudy site(4strata) andrandomly allocatedwithinstrata to one ofthe three study armsusing permuted blockrandomisation, withfixed block sizesof 3; thus, ineach ofthe four study sites,three children's centreswereallocated toeach arm. The allocation schedulewasproduced byan independent statistician,usingtheStata randomisa- tion algorithm. Allocations wereplaced insequentially numberedopaqueenvelopes (oneset for each trialsite). When astudy site(stratum) hadrecruited itsfirst three children's centres, and each ofthose centres hadrecruited 30families, thefirst three envelopes wereopened byan administrator nototherwise involvedinthe trial andtheallocation ofthose centres tostudy arm wasrevealed. Thesame process wasfollowed aftertherecruitment ofchildren's centres 4±6 and 5±9ateach study site.
Blinding It was notpossible toblind children's centrestaff,orresearchers providingtheintervention to treatment armallocation, butparents wereblinded totreatment armallocation. Analyseswere undertaken blindtotreatment armallocation.
Sample size The sample sizewas calculated forthe primary outcome (abinary variable forwhether ornot the family hadaplan forescaping fromahouse fire).Thenumber offamilies andclusters required wereobtained bycalculating thesample sizerequired foranindividually randomised trial then applying thedesign effectderived fromtheintraclass correlation coefficientandthe cluster sizetoaccount forclustering offamilies withinchildren's centresusingpublished for- mulae forcluster randomised trials.[27]Elevenchildren's centresperintervention arm(atotal of 33 centres acrossthefour study sites)wererequired todetect anabsolute difference inthe percentage offamilies withaplan forescaping fromahouse fireof20% (equivalent toan odds ratio of2.25) ineither ofthe two intervention armscompared withthecontrol arm.This assumed aprevalence of42% forfamilies inthe control armhaving aplan forescaping froma house fire,anintraclass correlation coefficientof0.05,[28] outcomes beingavailable on20 families perchildren's centre(giving adesign effectof1.95), 80%power and5%significance level (2-sided). Toallow forattrition, weincreased thenumber ofchildren's centresto36 and the average number offamilies recruited percentre to30, giving atotal of1080 families across the four study sites(9children's centresand270families perstudy site).
Statistical analysis Baseline characteristics weresummarised bytreatment arm.Quantitative analyseswereunder- taken usingapre-specified analysisplanonanintention-to-treat basisusing Stataversions 11 and 13.
Primary outcome. Theprimary outcome (abinary outcome ofwhether thefamily hada plan forescaping fromahouse fire)wasanalysed usingrandom effectslogistic regression to estimate oddsratios and95% CIs,comparing familiesinthe two intervention armswiththe control arm,withchildren's centreincluded asarandom effect.Themodel included Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 6/ 23 randomisation stratum(trialsite)asafixed effect andwasadjusted fortwo children's centre- level variables (leadagency (LocalAuthority, NationalHealthService orVoluntary sector)and Ofsted scoresforoverall effectiveness (Ofstedinspects andregulates servicesthatcare forchil- dren andyoung people andservices providing education andskills forlearners ofall ages) and two family-level variables(planforescaping fromahouse fireatbaseline andIndex ofMultiple Deprivation (IMD)2010score based onhome postcode).[16] TheIMD isasmall (400±1200 households) area-basedmeasureofdeprivation comprisingsevendomains (income, employ- ment, health anddisability, education skillsandtraining, barrierstohousing andservices, liv- ing environment andcrime).
As asecondary analysis,wetested fordifferential effectsofinterventions bydeprivation by adding interaction termstothe model. Weused oneway analysis ofvariance tocalculate the intraclass correlation coefficient.
Secondary outcomes. Forsecondary outcomesmeasured atfamily-level, weanalysed binary outcomes usingrandom effectslogistic regression, thenumber ofcorrect responses to fire safety knowledge questionsusingrandom effectsordinal regression andthebedtime safety routine scoreusing random effectslinearregression. Modelswereadjusted forthe same base- line covariates includedinthe model forthe primary outcome, asdescribed above.Statistical analysis ofsecondary outcomesmeasured atchildren's centre-level wasnotundertaken dueto small numbers insome groups.
Sensitivity analyses.Themain analyses foralloutcomes werecomplete caseanalyses. For the primary outcome weundertook sensitivityanalyses,(a)using multiple imputation to replace missing valuesandcreated 50datasets whichwerecombined usingRubin's rules,[29] and, (b)assuming nochange frombaseline valuesinthose losttofollow-up.
Health economic evaluation The cost-effectiveness analysiswasconducted fromasocietal perspective andused theprimary effectiveness endpointofthe trial andeconomic endpointofthe total costofthe intervention (expressed in2012 UK£), withdataanalysed atfamily-level. Resourceuseand cost data were obtained from:i)activity logs(relating toimplementation ofintervention); ii)children's centre follow-up questionnaires (detailingfiresafety activities); iii)parent follow-up questionnaires (resources andcosts related tofire safety sessions andhome safetyinspections). Studysiteand children's centre-level costswereaveraged equallyacrossfamilies withineachstudy siteand children's centre,respectively, andcombined withfamily-level coststogive atotal costper family. Theprimary outcome wascost peradditional planforescaping fromahouse fireesti- mated forthe IPB only andIPB+ armscompared tothe usual carearm.
We adopted ahierarchical modellingapproach,allowingforclustering andadjusting for the baseline covariates includedinthe primary effectiveness analysis.Thismodel extended recently developed methodology [30,31]forcost-effectiveness analysisalongside clustertrials (see S1Text forfurther details). Thisapproach usedMarkov ChainMonte Carlosimulation to fit the non-standard statisticalmodelusingtheWinBUGS software.[32] Asummary ofthe base-case cost-effectiveness analysisisprovided inTable 1.
One ofthe children's centresinthe usual caregroup hadextremely highcosts andits impact onthe results wasassessed byexcluding itin asensitivity analysis.Wechecked robust- ness offindings tomissing databyextending theimputation modeldescribed aboveforthe primary outcome toinclude costsincurred by(a) parents, (b)fire and rescue services, (c)chil- dren's centres and(d)other agencies. Sinceanalysis wascarried outusing MCMC simulation it was notpractical toperform 50imputations (asdone foreffectiveness), instead10imputa- tions wereused.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 7/ 23 Qualitative analysis Data from facilitation contactswereanalysed usingthematic analysisaftercategorisation into main sub-headings.[33] Ananalytical framework wasdeveloped andapplied toimplementa- tion fidelity interviews usingFramework Analysis,[34] supportedbyQSR NVivo 10.
Changes tothe protocol There weresixamendments madetothe protocol following approvalbythe ethics committee.
These were:
1. Specifying thetrial wasacluster randomised controlledtrialbyadding theword ªclusterº to the trial titleinthe protocol 2. Extending recruitment ofchildren's centrestothose withmore than50%oftheir catchment population inthe most 30%most deprived areastoensure recruitment ofsufficient chil- dren's centres 3. Increasing thenumber ofchildren's centremanagers interviewed inthe intervention arms (from asample of18 tomanagers fromall24 centres) toensure dataonfacilitators andbar- riers toimplementation ofthe IPB were collected fromallcentres 4. Reducing datacollection contactsbetween research teamandtheIPB only armtoasingle contact at12 months toreduce thelikelihood thatdata collection contactswouldactas prompts toaction toimplement guidancethatwould notusually occurwhenguidance doc- uments aredisseminated 5. The independent TrialSteering Committee (TSC)recommended modifyingthedata collec- tion atfacilitation contactsinthe `IPB plusfacilitation' armsothat atwo-stage approach was used, withaquestionnaire sentprior tothe interview toallow timeforthe data tobe collected, ratherthanthequestionnaire beingcompleted atthe interview.
6. The independent TSCadvised usinglatent variable analysisonthe baseline datatocreate a composite secondaryoutcomemeasurewhichmeasured arange offire escape behaviours and toinclude thisasan additional secondary outcomemeasureforthe trial.[35] Thiswas Table 1.Summa ryof the base-cas ecost-effective nessanalysis.
Type ofevaluation Prospectivecost-util ityanalysis alongsideacluster RCT Time horizon 1-year Perspecti ve Societal Comparat ors Usualcare Injury prevention brie®ng Injury prevention brie®ngwithfacilitat ion Cost categories Children'scentre Fire andRescue Service Other agencies includingcouncil Family Base yearforcalculatin g costs/pric es 2012 UK£ Analytic methods Hierarchicalmodel allowing forclustering andadjusting forthe baseli ne covariate sincluded inthe primary effectivene ssanalysis Outcome Costperadditio nal®re escape plan https://d oi.org/10.1371/j ournal.pone.0172584.t00 1 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 8/ 23 approved bythe ethics committee on7/2/13. Thiswasprior tocompletion offollow updata collection andprior toany analysis offollow updata.
Ethical andorganisational review Derbyshire ResearchEthicsCommittee (11/EM/0011) andtheUniversity ofthe West of England, Bristol,Research EthicsCommittee (HSC/11/06/61) providedethicalapproval on 18 th March 2011and22 nd July 2011, respectively. PrimaryCareTrusts (PCTs) provided NHS organisational approval.Informed writtenconsent wasobtained fromallparticipants involved.
Results Recruitment andretention Fig 1shows theflow ofchildren's centresandfamilies through thetrial. Ninety-six children's centres wereapproached, sevenofwhich wereexcluded. Fifty-seven centresexpressed interest in taking part,18ofwhich wereexcluded. Thirty-nine centreswererecruited, sixofwhich Fig 1.Flow ofchildren's centresandfamilies throughthe trial.
https://d oi.org/10.1371 /journal.pone.0172584.g001 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 9/ 23 were jointly operating asthree children's centres,givingatotal of36 centres. Atotal of1265 families wereapproached, ofwhom 1112were recruited. Allchildren's centresremained inthe trial and361families (32%)werelosttofollow-up. Losstofollow-up wassimilar acrosstrial arms.
Baseline characteristics Table 2shows thecharacteristics ofchildren's centresandfamilies atbaseline. Trialarms appear tobe well balanced. Twofifths offamilies (42%)hadanexisting planforescaping from a house firewith asimilar proportion ineach arm.
Primary andsecondary outcomes There wasnosignificant differenceinthe proportion offamilies withaplan forescaping from a house fireat12 months between treatment arms(AOR IPBonly vs.usual care:0.93,95%CI 0.58, 1.49;AOR IPB+ facilitation vs.usual care1.41, 95%CI 0.91,2.20) (Table 3;results forfull model inS1 Table). Therewasnosignificant interaction betweendeprivation andtheeffect of the interventions (p=0.86). Theintraclass correlation coefficientforhaving aplan forescap- ing from ahouse fireat12 months was0.003 (95%CI0.000, 0.027) (S2Table).
Significantly moreIPBonly (AOR 2.56,95%CI 1.38,4.76) andIPB+ armfamilies (AOR 1.78, 95%CI 1.01±3.15) wereinthe ªmore behaviours forescaping fromhouse firesºgroup than usual carearmfamilies. Families inthe IPB only armwere significantly lesslikely to report children playingwithmatches orlighters (AOR0.27,95%CI 0.08,0.94) andreported significantly morebedtime firesafety routines thanusual carearmfamilies (AORforaone unit increase innumber ofbedtime firesafety routines AOR1.59,95%CI 1.09,2.31).
Table 4shows asignificantly higherproportion ofIPB+ armfamilies reported receiving advice abouteachofthe five keyIPB messages andattended firesafety sessions oneach ofthe five keyIPB messages thancontrol armfamilies. Asignificantly higherproportion ofIPB only arm families attended firesafety sessions onthree ofthe five keyIPB messages thancontrol arm families. Asignificantly higherproportion offamilies inboth intervention armsreported attending afire safety session thancontrol armfamilies. Onlyasmall proportion offamilies attended twoormore firesafety sessions (usualcare:3.9%, IPBonly: 11.9%, IPB+19.3%).
There werenosignificant differences inother secondary outcomemeasures. Table5shows the fire safety activities reportedbychildren's centres.Numbers weretoosmall forstatistical anal- ysis butthefindings areconsistent withfamily-reported firesafety activities.
Sensitivity analyses AORs fromthemultiple imputation analysisforthe primary outcome weresimilar tothose from thecomplete caseanalysis (IPBonly: AOR 0.92,95%CI 0.58,1.46;IPB+: AOR1.40,95% CI 0.89, 2.21) andfrom theanalysis assuming nochange frombaseline (IPBonly: AOR 0.95, 95%CI 0.60,1.51;IPB+: AOR1.39,95%CI 0.91,2.12).
Health economics Details aboutderivation ofcostings arepresented inTables 6±9.Thecost ofdeveloping the IPB (£15,860) wasexcluded fromthecost-effectiveness analysisasthis fixed, one-off cost would notbeencountered againifthis intervention wasimplemented inpractice.
Table 10presents theresults ofthe cost-effectiveness analysiscomparing IPBonly andIPB+ to usual carearms. Thecomplete-case analysisshowstheIPB only isboth lesscostly andmore effective thanusual care,whereas theIPB+ ismore costly andmore effective thanusual care.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 10/ 23 Table 2.Baseline characteristicsofchildren's centresandfamilies.
TrialArm Characte ristics Usualcaren=12 IPBonly n=12 IPB+facilitation n=12 Children 'scentres Study centre:
Nottingha m 33 3 Newcastle 33 3 Norwich 33 3 Bristol 33 3 Lead agency:
Local Authorit y 107 9 NHS 00 2 Voluntary sector 25 1 Phase 1centre 1011 11 Phase 2centre 21 1 Number ofchildren incatchmen tarea: median (IQR) 754(529, 999) 776(565, 905) 854(608, 1076) Ofsted scoreforoverall effectivenes s: [1] Outstandin g 32 4 Good/sa tisfactory 910 7 Children's centreprovides adviceon:
Smoke alarmuse 1210 12 How tomake aplan forescaping fromahouse ®re 9[1]9 9 Causes ofhouse ®res 1211 12 Safe useand storage ofcigarettes ,lighters &matches 1010 [1]9 Bedtime routinestoprevent ®res 6[1]7 9 Families Usualcaren=370 (%) IPBonly n=369 (%) IPB+facilitation n=373 (%) Study centre:
Nottingha m 89(24.1) 98(26.6) 91(24.4) Newcastle 86(23.2) 88(23.9) 87(23.3) Norwich 95(25.7) 82(22.2) 93(24.9) Bristol 100(27.0) 101(27.4) 102(27.4) Single adulthouse hold [13]61(17.1) [15]72(20.3) [15]59(16.5) Only 1child inhouseh old [12]169(47.2) [13]173(48.6) [16]200(56.0) !1 family livinginsame househ old 29(8.0) 37(10.0) 37(10.0) Number offamilies withchildren aged: [10][6] [9] under 1year 163(45.3) 143(39.4) 178(48.9) 1±2 years 197(54.7) 220(60.6) 186(51.1) Mother aged 20years [16]17(4.8) [19]17(4.9) [17]20(5.6) Father aged 20years [57]6(1.9) [67]6(2.0) [59]8(2.6) Family ethnicity: WhiteBritish [20]337(96.3) [18]323(92.0) [12]348(96.4) English as®rst language [5]336 (92.1) [5]319 (87.6) [2]349 (93.1) Rented accomm odation [13]193(54.1) [6]203 (55.9) [6]193 (52.6) Deprivati onmean (SD)(IMD ofhouseh old) [2]31.0 (16.9) [1]34.7 (16.5) [1]29.6 (16.1) No smokers inhousehold [11]245(68.3) [9]251 (69.7) [10]263(72.5) At least oneperson inhousehold drinks 4times/wee k [10]21(5.8) [8]19(5.3) [12]24(6.7) At least oneperson inhousehold drinks 6drinks onone occasion [33]208(61.7) [38]173(52.3) [29]211(61.3) Family haveaplan forescaping fromahouse ®re [7]159 (43.8) [5]153 (42.0) [7]149 (40.7) [] = missing values https://do i.org/10.1371/j ournal.pone.0172584.t002 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 11/ 23 Table 3.Primary andsecondary outcomesat 12 month follow-upbytreatme ntarm.
Outcome measures TrialArm IPBonly vs.usual care IPB+vs.usual care Usual care n =258 (%) IPB only n =252 (%) IPB+ n=241 (%) Odds ratio (95% CI) p value Odds ratio (95% CI) p value Primary outcome measure 1 Family haveaplan forescaping fromahouse ®re: [4] [9][5] No 135(53.2) 135(55.6) 116(49.2) 1.00 1.00 Yes 119(46.9) 108(44.4) 120(50.9) 0.93(0.58, 1.49) 0.76 1.41(0.91, 2.20) 0.13 Second aryoutcome measures 2 Fire escape behaviours compositevariable:
Fewer ®reescape behaviours 45(17.4) 29(11.5) 32(13.3) 1.00 1.00 More ®reescape behaviours 213(82.3) 223(88.5) 209(86.7) 2.56(1.38, 4.76) 0.01 1.78(1.01, 3.15) 0.05 Smoke alarms®ttedandworking onevery level: [7] [12][8] No 22(8.8) 14(5.8) 13(5.6) 1.00 1.00 Yes 229(91.2) 226(94.2) 220(94.4) 1.61(0.71, 3.66) 0.25 1.56(0.71, 3.42) 0.27 Fire setting ormatch playbychildren: [52][49][49] No 197(95.6) 198(97.5) 181(84.3) 1.00 1.00 Yes 9(3.5) 5(2.5) 11(5.7) 0.27(0.08, 0.94) 0.04 1.22(0.43, 3.08) 0.77 Bedtime ®resafety routine score(median (IQR)):
3 [9] 8(8, 9) [16]9(8, 10) [11]8.5(8,9) 1.59(1.09, 2.31) 0.02 1.22(0.85, 1.76) 0.28 Took partinsmoking cessationcourses/sup port: [60] [43][43] No 5(19.2) 8(33.3) 5(23.8) 1.00 1.00 Yes 21(80.8) 16(66.7) 16(76.2) 0.23(0.04, 1.43) 0.12 0.61(0.11, 3.40) 0.57 Number ofcorrect responses to®re safety knowledge questions:
3 0 78(30.2) 81(32.1) 70(29.1) 1.10(0.77, 1.57) 0.61 1.22(0.86, 1.73) 0.26 1 93(36.1) 86(34.1) 76(31.5) 2 80(31.0) 81(32.1) 85(35.3) 3 7(2.7) 4(1.6) 10(4.2) Satisfaction withhome safety informati on provided bychildren's centre:
4 [57] [55][53] Neither satis®ednor dissatis®ed /fairly/very dissatis®e d 16 (8.0) 22(11.2) 23(12.2) 1.00 1.00 Very/fairly satis®ed 31(15.4) 46(23.4) 73(38.8) 1.08(0.4, 2.8) 0.871.79(0.7, 4.4) 0.20 No information received 154(76.6) 129(65.5) 92(48.9) [] = missing values 1 Adjusted forstudy centre, leadagency ofchildren's centre(Localauthority, NHSorVoluntary sector),OFSTED effectivene ssscore (Outsta nding,Good, Satisfactory ,Missing), ®reescape planatbaseline (no/yes), IMDscore offamily (continuo us).Wehave notadjusted forOFSTED capacityforsustained improvem entasthis variable hadmore missing dataandwhere recorded thevalues arethesame asfor OFSTED overalleffectivene ssscore.
2 Adjusted forthe lead agency ofthe children's centre,Ofstedreportscores foroverall effectivenes s,baseline valueofthe secondar youtcome measure, IMD. Wehave notadjusted forOFSTED capacityforsustaine dimprovemen tas this variable hadmore missing dataandwhere recorde dthe values arethe same asfor OFSTED overalleffectivenes sscore.
3 Odds ratioforaone unitincrease inthe outcome measure.
4 Participant swho hadnotreceived informatio nwere excluded fromtheanalysis https://do i.org/10.1371/j ournal.pone.0172584.t003 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 12/ 23 Table 4.Receipt offire safety advice andother firesafety promotio nby families atfollow-up, bytreatment arm.
Receipt of®re safety advice andpromotion TrialArm IPBonly vs.usual care IPB+ vs.usual care Usual care n =258 (%) IPB only n =252 (%) IPB+ n=241 (%) Odds ratio (95% CI) 1 p value Odds ratio (95% CI) 1 p value Received adviceonthe ®ve key IPB message s:
i. Smoke alarms 2 [54] [52][54] No 155(75.6) 132(66.0) 107(57.2) 1.00 1.00 Yes 49(24.0) 68(34.0) 80(42.8) 1.36(0.82, 2.26) 0.23 2.27(1.40, 3.67) p 0.01 ii. Matches 2 [56] [57][58] No 177(87.6) 167(85.6) 133(72.7) 1.00 1.00 Yes 25(12.4) 28(14.4) 50(27.3) 1.05(0.54, 2.04) 0.89 2.74(1.51, 4.96) p 0.01 iii. Fire escape plans [55][58][57] No 175(86.2) 168(86.6) 133(72.3) 1.00 1.00 Yes 28(13.8) 26(13.4) 51(27.7) 0.79(0.40, 1.55) 0.50 2.38(1.35, 4.21) p 0.01 iv. Bedtime safetyroutines [54][56][56] No 183(89.7) 173(88.3) 147(79.5) 1.00 1.00 Yes 21(10.3) 23(11.7) 38(20.5) 0.89(0.44, 1.82) 0.76 2.21(1.18, 4.12) p 0.01 v. Causes of®res [57][56][57] No 169(84.1) 149(76.0) 113(61.4) 1.00 1.00 Yes 32(15.9) 47(24.0) 71(38.6) 1.50(0.85, 2.65) 0.17 3.35(1.98, 5.68) p 0.01 Number ofkey safety message shad advice on: [53] [52][52] 2 or less 180(87.8) 170(85.0) 132(69.8) 1.00 1.00 3±5 25(12.2) 30(15.0) 57(30.2) 1.09(0.57, 2.10) 0.80 3.06(1.72, 5.43) p 0.01 Attended a®re safety session inthe last year: [53][50][49] No 197(96.1) 178(88.1) 155(80.7) 1.00 1.00 Attended 1or more 8(3.9) 24(11.9) 37(19.3) 3.20(1.27, 8.06) 0.01 7.07(3.05, 16.38) p 0.01 Attended a®re safety session atchildre n'scentre: [53][50][49] No 197(96.1) 185(91.6) 163(84.9) 1.00 1.00 Attended 1or more 8(3.9) 17(8.4) 29(15.1) 2.18(0.85, 5.63) 0.11 5.14(2.20, 12.03) p 0.01 Attended ®resafety session abouteachofthe ®ve key message sin the IPB inthe last year: [53] [50][49] i. Smoke alarms No 198(96.6) 180(89.1) 158(82.3) 1.00 1.00 Yes 7(3.4) 22(10.9) 34(17.7) 3.34(1.30, 8.58) 0.01 6.71(2.80, 16.04) p 0.01 ii. Matches No 201(98.1) 189(93.6) 169(88.0) 1.00 1.00 Yes 4(2.0) 13(6.4) 23(12.0) 2.80(0.85, 9.29) 0.09 6.78(2.24, 20.55) p 0.01 iii. Fire escape plans No 201(98.1) 188(93.1) 162(84.4) 1.00 1.00 Yes 4(2.0) 13(6.9) 30(15.6) 3.48(1.06, 11.44) 0.04 9.88(3.31, 29.43) p 0.01 iv. Bedtime safetyroutines (Continued) Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 13/ 23 Table 4.(Continue d) Receipt of®re safety advice andpromotion TrialArm IPBonly vs.usual care IPB+ vs.usual care Usual care n =258 (%) IPB only n =252 (%) IPB+ n=241 (%) Odds ratio (95% CI) 1 p value Odds ratio (95% CI) 1 p value No 202(98.5) 189(93.6) 172(89.6) 1.00 1.00 Yes 3(1.5) 13(6.4) 20(10.4) 3.93(1.04, 14.93) 0.04 7.83(2.23, 27.55) p 0.01 v. Causes of®res No 198(96.6) 184(91.1) 162(84.4) 1.00 1.00 Yes 7(3.1) 18(8.9) 30(15.6) 0.56(0.0, 11.9) 0.06 5.52(2.29, 13.30) p 0.01 [] = missing Some families attended morethan1session .
1 Adjusted forstudy centre, leadagency ofchildren's centre(Localauthority, NHSorVoluntary sector),OFSTED effectivene ssscore (Outsta nding,Good, Satisfactory ,Missing), baselinevalueofthe secondary outcomemeasure, IMDscore offamily (continuous ).We have notadjusted forOFSTED capacityfor sustained improvementasthis variabl ehad more missing dataandwhere recorded thevalues arethesame asfor OFSTED overalleffectiven essscore.
2 IMD quintiles usedbecause ofnon-linear association withtheoutcome.
https://do i.org/10.1371/j ournal.pone.0172584.t004 Table 5.Fire safety activitiesreported bychildren's centresatfollow-up ,by treatmen tarm.
Second aryoutcome measures Trialarm Usual care IPBonly IPB+facilitation n =12 n=12 n=12 Advice provided on:
Smoke alarms [1] No advice /Don't know 10 1 Yes 1111 11 How tomake a®re escape plan [1] No advice /Don't know 33 0 Yes 89 12 Causes ofhouse ®res(cooking safety,electrica lsafety, handlin ghot irons safely) No advice /Don't know 10 0 Yes 1112 12 Child behaviour and®repreven tion(safe useand storage ofcigarettes ,lighters andmatches ) No advice /Don't know 411 0 Yes 81 12 Bedtime routinestoprevent ®res [1][1] No advice /Don't know 64 0 Yes 58 11 Children's centreprovided ®resafety session s [1] No 65 1 Yes- 57 11 Mean number ofsession s(min tomax) 1.2(1to2) 2.1(1to4) 3.1(1to7) Mean session lengthinminutes (mintomax) 116(90to120) 90(30 to120) 89(30 to130) FRS attended tohelp provide anysessions [1][1] No 86 4 yes 36 7 [] missing values.
assumes thosewhosaidthey ranasession butdidn't answer question onnumber ofsessions, ranonly onesession.
FRS =®re and rescue service https://do i.org/10.1371/j ournal.pone.0172584.t005 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 14/ 23 As the inverse ofthe difference inprobabilities ofhaving aplan forescaping fromahouse fire between twotreatment armsisequal tothe Number NeededtoTreat (NNT), thecost-effective- ness ratios canbeinterpreted asthe cost peradditional planforescaping fromahouse fire under theintervention. Thecost-effectiveness acceptabilitycurves(Fig2)show theIPB only has thehighest probability ofbeing cost-effective atadecision maker'swillingness topay between £0and £4,000 (US$6061) peradditional planforescaping fromahouse fire.
A sensitivity analysis,excluding thechildren's centrewiththepotentially outlyingcost, resulted inslight changes asreported inTable 10and Fig2.Analysis ofimputed data,atvalues of adecision maker'swillingness topay above £400(US$ 606)peradditional planforescaping from ahouse fire,showed theIPB+ hadthehighest probability ofbeing cost-effective (Table 10)reaching aprobability ofnearly 1at £1000 (US$1515) peradditional fire-escape plan. However, resultsshould beinterpreted withcaution duetothe large proportion of Table 6.Sources ofunit cost data(UK£ 2012).
Value Source Parent costs Time costs £45.70/hourDepartmentfor Transport. TAGUNIT 3.5.6Values ofTime andVehicle Operating CostsTransport Analysis Guidanc e(TAG) .Available fromwww.dft .gov.uk/web tag[Accessed October2012].
Travel costsbycar £0.18/km Departmentfor Transport. TAGUNIT 3.5.6Values ofTime andVehicle Operating CostsTransport Analysis Guidanc e(TAG) .Available fromwww.dft .gov.uk/web tag[Accessed October2012].
IPB implemen tationcosts Research er'stime £19.04/hour Universityof Nottingham payscale Administra tor'stime £11.24/hour Universityof Nottingham payscale Children 'scentre, FireandRescue Serviceandother agency costs FRS stafftime £36.00/hour Personalcommun ication,AdamShaw, Cheshire FireandRescue Service, 20September 2012, Children's centrestaff's time £18.00/hour PersonalSocial Services Research Unit.Unitcosts ofhealth andsocial care2012, 2012.(assumed sameas home careworker) Home inspection £15.33Basedon40minute visitbychildre n'scentre, FRSorother agency (asindecision models) FRS =®re and rescue service https://do i.org/10.1371/j ournal.pone.0172584.t006 Table 7.Costs ofproviding theIPB, trainin gand facilit ation (UK£2012).
Arm Study centre Number of families random ised Number of childr en's centres IPB printing & distributio nIPB training session IPB facilitatio nTotal per study centre Total per childr en's centre Total perfamily randomi sed Usual care Bristol 1003£0.00 £0.00£0.00 £0.00£0.00 £0.00 Newcastle 863£0.00 £0.00£0.00 £0.00£0.00 £0.00 Norwich 953£0.00 £0.00£0.00 £0.00£0.00 £0.00 Nottingham 893£0.00 £0.00£0.00 £0.00£0.00 £0.00 IPB only Bristol 1013£152.50 £0.00£0.00£152.50 £50.83 £1.51 Newcastle 883£152.50 £0.00£0.00£152.50 £50.83 £1.73 Norwich 823£152.50 £0.00£0.00£152.50 £50.83 £1.86 Nottingham 983£152.50 £0.00£0.00£152.50 £50.83 £1.56 IPB + Bristol 1023£152.50 £1,328.95 £327.84£1,809.29 £603.10 £17.74 Newcastle 873£152.50 £1,408.84 £220.57£1,781.91 £593.97 £20.48 Norwich 933£152.50 £1,488.74 £127.90£1,769.14 £589.71 £19.02 Nottingham 913£152.50 £1,568.63 £84.74£1,805.87 £601.96 £19.84 https://do i.org/10.1371/j ournal.pone.0172584.t007 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 15/ 23 missing dataimputed (ranging fromjustunder 50%(families' costs)tonearly 60%(children's centre costs)).
Qualitative analysis The qualitative datahave been reported previously.[14] However,theresults provided clear indications offactors, e.g.organisational change,timeandresources whichmoderated allchil- dren's centres' abilitytoimplement theIPB. More specific factorswhichadversely affected implementation ofthe IPB bysome children's centreswerealsoidentified, e.g.staff training and continuity.
Discussion Main findings This three-arm clusterrandomised controlledtrialexamined theeffects ofacomplex interven- tion onfire safety behaviours inthe home. Wefound thatfamilies inboth intervention arms reported significantly morebehaviours forescaping fromhouse fires.Families inboth Table 9.Total interventio ncosts expressed percluster (i.e.children 'scentre) andperfamily.
Usual care-Mean (MintoMax) IPBonly -Mean (MintoMix) IPB+facilitation -Mean (Minto Max) per cluster perfamily percluster perfamily percluster perfamily 9 cluste rs 151families 9clusters 140families 9clusters 123families IPB provision ,training andfacilitat ion plus other interventi oncosts £303.01 (30.66to 1367.67) £18.06 (0.00to 90.43) £143.68 (62.51to 240.22) £9.24 (1.51to 46.79) £507.81 (290.26 to 859.92) £37.16 (21.18to 79.20) Other interventio ncosts only £303.01(30.66to 1367.67) £18.06 (0to 90.43) £117.87 (45.89to 210.58) £7.58 (0.00to 45.06) £224.97 (110.66 to 497.62) £16.46 (2.70to 51.33) https://do i.org/10.1371/j ournal.pone.0172584.t009 Table 8.Other intervention costsexpressed percluster (i.e.children 'scentre) andperfamily.
Usual care-Mean (MintoMax) IPBonly -Mean (MintoMix) IPB+facilitation -Mean (MintoMax) per cluster perfamily percluster perfamily percluster perfamily 9 clusters 151families 9clusters 140families 9clusters 123families Fire safety sessions Children's centrecosts £421.00(0.00to 1800 ) £13.72 (0.00to 62.07) £63.00 (0.00to 198.00) £2.06 (0.00to 7.07) £222.00 (0.00to 900.00 ) £7.63 (0.00to 32.14) Fire &Rescue service costs £74.38 (0.00to 378.00) £2.57 (0.00to 13.03) £68.66 (0.00to 288.00) £2.20 (0.00to 9.60) £136.49 (0.00to 372.00 ) £4.38 (0.00to 11.63) Parent coststoattend sessions £2.59 (0.00to 22.85) £0.15 (0.00to 22.85) £9.15 (0.00to 22.85) £0.59 (0.00to 17.14) £14.83 (0.00to 62.32) £1.09 (0.00to 35.18) Home safety inspectio ns Children's centrecosts £3.41(0.00to 15.33) £0.20 (0.00to 15.33) £6.81 (0.00to 30.66) £0.44 (0.00to 15.33) £8.52 (0.00to30.66) £0.62(0.00to 15.33) Fire &Rescue Service costs £13.63 (0.00to 30.66) £0.76 (0.00to 15.33) £13.63 (0.00to 30.66) £1.07 (0.00to 15.33) £22.14 (0.00to 76.65) £1.62 (0.00to 15.33) Other agencies £10.22(0.00to 45.99) £0.61 (0.00to 15.33) £22.14 (0.00to 45.99) £1.42 (0.00to 45.99) £15.33 (0to45.99) £1.12(0.00to 15.33) Complete dataoncosts available for9clusters perarm. Thenumber offamilies represen tsthe number offamilies withcomplete datawithin the9clusters per arm.
Possible outlier(reporte d20 children's centrestaff+5Fire &Rescue Servicestaffproviding aªfun dayº (cluster 6,usual carearm))Ðw henthiscluster is removed themean isreduced to£254.25 withamaximum percluster of£792.00 .
https://do i.org/10.1371/j ournal.pone.0172584.t008 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 16/ 23 intervention armsweresignificantly morelikely toreport attending afire safety session than usual carearmfamilies. However, interms ofthe primary outcome measurefamiliesineither intervention armwere notsignificantly morelikely toreport having aplan forescaping froma house firethan usual carearmfamilies. Economic analysisshowedtheIPB only intervention was both lesscostly andmore effective thanusual care,whereas theIPB +facilitation was more costly butalso more effective thanusual care.
Strengths andlimitations Our trialused atheoretically-based intervention.Itmeasured awide range ofoutcome and process measures andused quantitative andqualitative approaches. Thisprovided agood understanding ofwhat theintervention comprisedofand how itmay have achieved itsimpact.
Recruitment exceededourrequired samplesize;theattrition ratewasconsistent withthe Table 10.Cost-effect ivenessanalysis resultsforcomplete caseandsensitiv ityanalyses .
Usual care IPBonly IPB+facilitati onIPBonly vs.
Usual care IPB +facilitatio n vs. Usual care Complete caseanalysis Number offamilies 151140 123 Number ofchildre n'scentres 99 9 Mean costperfamily (95%Credib leinterval (CrI)) £21.15(3.95to 38.31) £12.65 (4.66to 20.03) £41.41 (31.58to 52.41) -£8.49 £20.26 Proportion withFireescape plan(95% CrI) 0.48(0.35 to 0.56) 0.49 (0.38 to 0.58) 0.48 (0.37 to0.58) 0.03 0.02 Increment alcost effectivenes sratio -£275.31 £1007.96 Probabilit ycost effective @£200 (US$ 306)peradditional plan forescaping fromahouse ®re 0.19 0.78 0.02 Probabilit ycost effective @£1000 (US$1531) per additional planforescaping fromahouse ®re 0.22 0.53 0.25 Sensitivity analysisomitting outlyingcluster Number offamilies 151140 123 Number ofchildren's centres 99 9 Mean costperfamily (95%CrI) £14.99(6.16to 24.11) £13.26 (4.52to 22.24) £39.97 (31.25to 48.41) -£1.74 £24.98 Proportion withplan forescaping fromahouse ®re(95% CrI) 0.47 (0.34 to 0.61) 0.50 (0.37 to 0.64) 0.48 (0.34 to0.62) 0.03 0.01 Increment alcost effectivenes sratio -£53.01£3778.55 Probabilit ycost effective @£200 (US$ 306)peradditional plan forescaping fromahouse ®re 0.33 0.63 0.03 Probabilit ycost effective @£1000 (US$1531) per additional planforescaping fromahouse ®re 0.27 0.53 0.20 Sensitivity analysisimputing formissing values Number offamilies 370369 373 Number ofchildren's centres 1212 12 Mean costperfamily (95%CrI) £19.21(14.64 to 23.77) £10.60 (6.48to 14.73) £43.01 (38.71to 47.30) -£8.60 £23.80 Proportion withplan forescaping fromahouse ®re(95% CrI) 0.44 (0.37 to 0.52) 0.44 (0.37 to 0.52) 0.58 (0.50 to0.65) -0.00 0.13 Increment alcost effectivenes sratio £6447.53 £177.61 Probabilit ycost effective @£200 (US$ 306)peradditional plan forescaping fromahouse ®re 0.02 0.60 0.24 Probabilit ycost effective @£1000 (US$1531) per additional planforescaping fromahouse ®re 0.00 0.02 0.96 https://do i.org/10.1371/j ournal.pone.0172584.t010 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 17/ 23 sample sizecalculation andwassimilar acrosstreatment arms.Themajority ofparticipants were from themost disadvantaged areashence ourintervention wasdelivered tofamilies at higher riskofinjury.
In terms oflimitations, fewparticipants camefromablack orethnic minority grouporhad English asasecond language. Outcomemeasureswereself-reported andalthough wewere able toblind parents totreatment armallocation, children'scentrestaffcould notbeblinded; hence thismay have influenced reportingofsome secondary outcomemeasures bychildren's centres. Furtherworkisrequired toexplore thepossibility thattheintervention increasedfam- ilies' understanding andhence theirreporting ofplans forescaping fromhouse fires.Thetrial had three armsandmultiple secondary outcomemeasures leadingtomultiple significance testing, although wedid pre-specify asingle primary outcome measure. Wedidnot adjust sig- nificance levelstoaccount forthe three armdesign orfor having multiple secondary outcomes, since there isno consensus onthis and ithas been stated thatformal adjustments formultiplic- ity usually complicate ratherthanenlighten.[36, 37]The significant resultsforsecondary out- comes should however beinterpreted withcaution. Finally,somefamilies didnot receive the intervention anditis possible, therefore, thatgreater implementation mayhave achieved Fig 2.Cost-ef fectiveness acceptabilitycurves forthe complete caseanalysis andwithout theoutlier children' scentre.
https://d oi.org/10.1371 /journal.pone.0172584.g002 Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 18/ 23 greater behavioural change.Arecent national evaluation ofchildren's centresinthe UK found most children's centreservices wereusedbyfamilies forless than oneyear.[38] Underthese conditions itis difficult toachieve highlevels ofpenetration ofinterventions.
Comparisons withexisting research The baseline prevalence ofplans forescaping fromhouse firesinour study wassimilar tothat in the US(52%), suggesting similarscopeforimprovement asin the UK. Wewere unable to find anypublished evaluations ofinjury prevention interventions deliveredinchildren's cen- tres inthe UK with which tocompare ourfindings. SureStart Local Programmes (SSLPs)were the forerunners tochildren's centres.Thenational evaluation ofSSLPs compared outcomesin SSLP families withthose inthe Millennium CohortStudylivinginsimilarly disadvantaged areas without SSLPs.[39] Atthe age ofthree years, children inSSLP areashadasignificantly lower unintentional injuryratethan those inthe non-SSLP areas;thisdifference wasnotmain- tained bythe time children reachedfiveyears ofage. Consistent withourfindings, thissug- gests suchservices canimpact onhome safetyoutcomes forfamilies indisadvantaged areas.
[40] There havebeen fewevaluations ofHead Startprogrammes inthe USwith injury preven- tion asthe focus eventhough children inthe programmes areknown tobe atrisk ofinjury.
[41, 42] Children's centresinthe UK have been evaluated inamulti-component sixyear study (The Evaluation ofChildren's CentresinEngland (ECCE)). Asurvey of5,700 parents participating in ECCE [38]suggested few(8%) hadreceived homesafetyadvice fromchildren's centres.The challenges ofdelivering evidence-based programmeswithinchildren's centreswereexplored in questionnaires andinterviews withstaffin121 children's centres.[43] Widespread useof evidence-based programmes,particularlyparentingprogrammes, wasfound. Education and training areimportant, notonly forparents, butalso forthose whocareforchildren andinflu- ence children.[2] Byintroducing theIPB tochildren's centrestaffboth thestaff andparents had exposure toevidence-based messages.However,asthe ECCE studyfound, somechildren's centre staffªgave equalweight toresearch evidence andpersonal experienceº andonly asmall number offamilies werereached bythe best evidenced programmes.[43] Similarly,wefound fewer than50%offamilies received eachkeysafety message andfewer than20%attended fire safety sessions.
Conclusion andrecommendatio ns This research hasdemonstrated thatchildren's centrescaneffectively deliverevidence-based injury prevention tofamilies withyoung children wholiveindisadvantaged communitiesand achieve changes insome home safetybehaviours. Theeconomic analysissuggested theIPB only appeared tobe most cost-effective, butthis should beinterpreted withcaution asthe results weresensitive tomissing data.
Choosing outcomemeasures forfire prevention programmes isdifficult. Fire-related inju- ries areuncommon events,andanextremely largetrialwould berequired toevaluate the impact ofan educational intervention onreducing fire-related injuries.Thereisevidence that some firesafety behaviours doreduce therisk offire-related injuriesanddeaths, suchas smoke alarms.[5, 6]However, smokealarmownership isvery common inthe UK (more than 90% offamilies inour trial reported afitted andworking smokealarmonevery leveloftheir home atbaseline), sodemonstrating increasesinfunctional ownership againrequires large sample sizes.Furthermore, thosewhodon't already havesmoke alarms maybeparticularly resistant tochange, makingonlysmall effect sizesfeasible todetect. Weused fireescape plans as the primary outcome measureforour trial andalso measured component elementsoffire Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 19/ 23 escape plansasasecondary outcomemeasure. Ourfinding ofsignificant increasesinthe com- ponent elements ofaplan butnot infire escape plansperse,suggests furtherworkisneeded to develop avalid andreliable measure offire escape plans.Research demonstrating areduc- tion inrisk offire-related injuryassociated withhaving afire escape planwould behelpful, as would largertrialsallowing detection ofsmaller, butclinically important differences.
The mode ofintervention, throughchildren's centrestaffrather thandirectly tofamilies, is typical ofmany health promotion programmes inpractice. Ourtrialsuggests lessthan full implementation ofthe IPB, hence future studies shouldconsider additional oralternative implementation strategies.
What isalready knownonthis subject?
Little isknown abouteffective measures toreduce therisk offire-related deathandinjury in children otherthanprovision ofsmoke alarms.
Children's centresinthe UK and Head Startinthe USboth have injury prevention asone of their priorities.
Educational interventions canincrease theprevalence ofplans forescaping fromhouse fires although theeffectiveness ofchildren's centresindelivering suchpackages isunknown.
What thisstudy adds?
An Injury Prevention Briefing(anevidence-based educationalpackagedelivered bychildren's centres) canimprove someaspects offamilies' behaviours forescaping fromhouse fires,e.g., making sureexits areclear.
The IPBonly intervention waslesscostly andmore effective thanusual care,whereas the IPB +facilitation wasmore costly butalso more effective thanusual care.These results open the way forthe development andevaluations ofInjury Prevention Briefingsforother typesof injury.
Trial registration Trial registration: ClinicalTrials.gov identifier:NCT01452191. Dateofregistration: 10/10/ 2011. ISRCTN trialidentifier: ISRCTN65067450. Dateofassignation: 06/12/2012.Thefirst children's centrewasrecruited tothe trial on3/8/2011. Thefirst parent wasrecruited tothe trial on5/1/2012. Trialregistration occurredafterrecruitment ofsome children's centresas the original protocol approved bythe ethics committee didnot include theword ªclusterº in the title andwewished toregister thetrial with thisspecified inthe title. This required aproto- col amendment andethical approval forthe amendment.
Supporting information S1 Table. Fullmodel forprimary outcome (familyhaveaplan forescaping fromahouse fire) at12 months follow-up.
(DOCX) S2 Table. Analysis ofvariance tableforprimary outcome (familyhaveaplan forescaping from ahouse fire)at12 months follow-up according toclustering bychildren's centre.
(DOCX) S1 Text. Technical appendix.
(DOCX) Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 20/ 23 S2 Text. KCSNon-IMP interventional trialprotocol V1.
(DOCX) S3 Text. PLoS ONECONSORT extensionforcluster trialschecklist.
(DOCX) Acknowledgmen ts The authors wishtothank theparticipating children'scentrestaffandmanagement andfami- lies inBristol, Newcastle, NorwichandNottingham areasfortheir support inundertaking this research. Withouttheirhelpthistrial would nothave been possible.
The views expressed arethose ofthe authors andnotnecessarily thoseofthe NHS, the NIHR orthe Department ofHealth.
Previously presented Multicentre randomised controlledtrialevaluating implementation ofafire-prevention injury prevention briefinginchildren's centres.PublicHealth Science; EUPHA Conference 7th European PublicHealth Conference, 19±22November 2014.
Author Contributions Conceptualization: DKRREM MH MW CCNC AS.
Data curation: AKCCDK TDTGKB.
Formal analysis: TDAH AKNC CCASGMN LMTGKB.
Funding acquisition: DKRREM MH MW CCNC AS.
Investigation: TDAH JAGMN LMTGKB.
Methodology: DKTDAH MH NCCCASGMN TGMW EM.
Project administration: JAAH GMN TG.
Software: NCAS.
Supervision: DKRREM TDCC.
Validation: KBTG.
Visualization: AKTDCCDK.
Writing ±original draft:DKTDNC ASMW JACC GMN.
Writing ±review &editing: DKTDAH AKMH NCMW JACC ASGMN LMTGKBEM RR.
References 1. Peden M,Oyegbite K,Ozanne-Sm ithJ,Hyder A,Branche C,Rahman A,etal. World Report onChild Injury Preven tion.World Health Organisatio nand UNICEF, Geneva,2008.
2. National CenterforInjury Prevention andControl. National ActionPlanforChild Injury Prevention: An Agenda toPreven tInjuries andPromote theSafety ofChildren andAdolesce ntsinthe United States.
Atlanta (GA):National CenterforInjury Preven tionand Control, 2012.
3. Sethi D,Towner E,Vincenten J,Segui-Gom ezM, Racioppi F.European ReportonChild Injury Preven- tion. Geneva ,Switzerlan d:World Health Organisatio n,2008.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 21/ 23 4. Edwards P,Roberts I,Green J,Lutchmun S.Deaths frominjury inchildren andemployment statusin family: analysis oftrends inclass specific deathrates. BMJ.2006; 333(7559) :119±21.https://doi.org/ 10.1136/ bmj.38875. 757488.4F PMID:168295 37 5. Runyan CW,Bangdiw alaSI,Linzer MA,Sacks JJ,Butts J.Risk factors forfatal residential fires.NEngl J Med. 1992; 327(12):85 9±63.https://doi.or g/10.1056/NEJM199209 173271207PMID: 1508246 6. Marshall SW,Runyan CW,Bangdiw alaSI,Linzer MA,Sacks JJ,Butts JD.Fatal residential firesÐWho dies andwho survives? JAMA.1998;279(20):16 33±7.PMID: 9613913 7. Kendrick D,Young B,Mason-Jo nesAJ,Ilyas N,Achana FA,Cooper NJ,etal. Home safety education and provision ofsafety equipme ntfor injury prevention. CochraneDatabaseSyst Rev. 2012; 9:
CD005014.
8. DiGuisep piC, Goss CW,Higgins JP.Interventi onsforpromoting smokealarmownership andfunction.
Cochrane DatabaseSystRev2001. 2:CD00 2246.
9. Hayes M,Kendrick D,Deave T.Injury prevention briefing.Preventing unintentional injuriestothe under fives: aguide forpractition ers.http://tiny.c c/kcspage2014.
10. Department forEducatio n.Sure Start Children's CentresStatutory Guidance.London: Department for Educatio n,2010.
11. Department forEducatio n.The Core Purpose ofSure Start Children 'sCentres: Governm entVision.
London: DepartmentforEducation, 2012.
12. Ramey SL,Ramey CT,Phillips MM,Lanzi RG,Brezause kC, Katholi CR,etal. Head StartChildren 's Entry intoPublic School: AReport onthe National HeadStart/ Public EarlyChildhood TransitionDem- onstratio nStudy. Washingto nDC, USA: USDept ofHealth andHuma nServices, 2000.
13. Tarullo LB,Doan H,Zill N,Westat I,Associates E.Head StartProgram performanc emeasures: Second progress report.Washingto nDC: Administra tionforChildren Youth,Families, 1995.
14. Beckett K,Goodenoug hT, Deave T,Jaeckle S,McDaid L,Benford P,etal. Implementin gan Injury Pre- vention Briefing toaid delivery ofkey firesafety messages inUK children's centres:qualitative study nested withinamulti-centr erandom isedcontrolled trial.BMC public health. 2014;14:1256 .https://doi.
org/10.1186/ 1471-2458-14-1256PMID:25492496 15. Deave T,Towner E,McColl E,Reading R,Sutton A,Coupland C,etal. Multicentre clusterrandomise d controlled trialevaluatin gimplementa tionofafire prevention InjuryPreven tionBriefing inchildren's cen- tres: study protocol .BMC public health. 2014;14(69).
16. Department forCommun itiesandLocal Governm ent.English IndicesofDeprivatio n2010. https://www.
gov.uk/gov ernment/sta tistics/englis h-indices-of- deprivation-2010 [Accessed 12/12/14].
17. Hindmarch P,Hawkins A,McColl E,Hayes M,Majsak-New manG,Ablewhi teJ,et al. Recruitment and retention strategiesand theexaminat ionofattrition biasinarandom isedcontrolled trialinchildren's centres servingfamilies indisadvanta gedareas ofEngland. Trials.2015;16(79).
18. Medical Research Council.Developing andevaluatin gcomplex interventions:newguidance .London:
2008.
19. Ingram JC,Deave T,Towner E,Errington G,Kay B,Kendrick D.Identifying facilitatorsand barriers for home injurypreventi oninterven tionsforpre-school children:asystematic reviewofthe quantita tiveliter- ature. Health EducRes.2012; 27(2):258±68 .Epub 26August 2011.https://doi.or g/10.1093/her/cyr06 6 PMID: 218736 13 20. KellyMP,Chambers J,Huntley J,Milward L.Evidence intopractice: Method1for the production of Effective ActionBriefings andrelated materials. http://www. nice.org.uk/aboutnice/whoweare/ aboutthe hda/evidenc ebase/keypap ers/evidenceintopratice/eviden ce_into_prac tice_method_1_for_ the_produc tion_of_effec tive_action_briefings_an d_related_mate rials.jsp[accessed 19/5/14].London:
Health Development Agency,2004.
21. Durlak JA,DuPre EP.Implementati onMatters: AReview ofResearch onthe Influence ofImplemen ta- tion onProgram Outcomes andtheFactors Affecting Implementati on.Am JCommun ityPsychol. 2008; 41:327±50. Epub6March 2008.https://doi.or g/10.1007/s10464-008- 9165-0PMID:1832279 0 22. StithS,Pruitt I,Dees J,Fronce M,Green N,Som A,etal. Implement ingCommunity -BasedPrevention Programm ing:AReview ofthe Literature. TheJournal ofPrimary Prevention. 2006;27(6):599± 617.
Epub 19October 2006.https://doi.or g/10.1007/s10935-006- 0062-8PMID:17051431 23. Fleuren M,Wiefferink K,Paulussen T.Determinan tsof innovation withinhealth careorganiz ations:Lit- erature reviewandDelphi study.Interna tionalJournal forQuality inHealth Care.2004; 16(2):107± 23.
https://doi.or g/10.1093/intqhc/mz h030PMID: 15051705 24. Greenhal ghT,Robert G,Bate P,Kyriakidou O,Macfarlan eF, Peacock R.How tospread goodideas: A systematic reviewofthe literature ondiffusion, disseminat ionand sustainabi lityofinnovatio nsinhealth service delivery andorganisation. http://www.nets.nih r.ac.uk/__da ta/assets/pdf_file/0017/ 64340/FR- 08-1201-0 38.pdf[Accessed10/12/15 ].2004.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 22/ 23 25. Gielen A,Sleet Da,DiClemente R.Injury andViolence Prevention: BehavioralScience Theroies ,Meth- ods and Applications. SanFrancisc o,CA: Jossey- Bass;2006.
26. Brussoni M,Towner E,Hayes M.Evidence intopractice: combining theartand science ofinjury preven- tion. InjPrev. 2006; 12(6):373± 7.https://doi.o rg/10.1136/ip .2005.011403PMID: 17170184 27. Eldridge S,Kerry S.APractical GuidetoCluster Randomis edTrials inHealth Services Research:
Wiley; 2012.Pages 141±2: Formulae 7.2and 7.5.
28. Deave T,Goodenou ghT,Stewart J,Towner E,Majsak-N ewmanG,Hawkins A,etal. Contempora ry hazards inthe home: keeping childrensafefrom thermal injuries. ArchDisChild. 2013;98:485±9.
https://doi.or g/10.1136/archdisch ild-2012-302901PMID: 2359272 9 29. Rubin DB.Multiple Imputation forNonrespon seinSurveys. NewYork: JohnWiley &Sons, 2004.
30. Grieve R,Nixon R,Thomps onS.Bayesian hierarchicalmodels forcost-effectiv enessanalyses thatuse data from cluster random izedtrials. Medical DecisionMaking. 2010;30(2):163± 75.Epub 2009Aug12.
https://doi.or g/10.1177/0272989X0 9341752PMID: 19675321 31. Gomes M,Grieve R,Nixon R,Ng ESW, Carpenter J,Thompson SG.Methods forcovariate adjustment in cost-effect ivenessanalysis thatusecluster random isedtrials. Health Economics .2012; 21(9):1101 ± 18. https://doi. org/10.1002/h ec.2812PMID:22461149 32. LunnD,Thomas A,Best N,Spiegelhal terD.WinBUG SÐABayesian modellingframework: Concepts, structure, andextensibili ty.Statistics andComputing. 2000;10(4):325± 37.
33. PopeC,Ziebland S,Mays N.Analysing qualitative data.BMJ. 2000; 320(7227) :114±6.PMID:
10625273 34. GaleNK,Heath G,Cameron E,Rashid S,Redwo odS.Using theframework methodforthe analysis of qualitativ edata inmulti-discip linaryhealth research .BMC Research Methodology.2013; 13(1):1±8.
35. Ploubidis GB,Edward sP, Kendrick D.Measurin gbehaviou rsfor escaping fromhouse fires:useof latent variable modelstosummarise multiplebehaviours. BMCResea rchNotes. 2015;8(1):1±8.
36. Schulz KF,Grimes DA.Multiplicity inrandom isedtrials I:endpoi ntsand treatments .The Lancet. 365 (9470):159 1±5.
37. Pocock SJ.Clinical trialswithmultiple outcomes: Astatistic alperspective ontheir design, analysis, and interpretatio n.Controlle dClinical Trials.1997;18(6):530 ±45.PMID: 9408716 38. Maisey R,Speight S,Haywood S,Hall J,Sammons P,Hussey D,etal. Evaluation ofChildren's Centres in England (ECCE). Strand2:Baseline SurveyofFamilies UsingChildren's Centresinthe Most Disad- vantaged Areas.Darlingto n:Department forEducatio n,2013.
39. NESS Researc hTeam. Theimpact ofSure Start Local Programm eson three yearoldsandtheir fami- lies. London :DCSF, 2008.
40. NESS Researc hTeam. Theimpact ofSure Start Local Programm eson five year oldsandtheir families.
London: 2010.
41. Johnston BD,Britt J,D'Ambrosio L,Mueller BA,Rivara FP.Apreschool programforsafety andinjury prevention deliveredbyhome visitors. InjPrev. 2000; 6(4):305±9 .https://doi.or g/10.1136/ip.6.4.305 PMID: 111446 34 42. Whiteside -MansellL,Johnson D,Aitken M,Bokony P,Conners- BurrowN,McKelvey L.Head Startand Unintende dInjury: TheUse ofthe Family MapInterview toDocument Risk.Early Childhoo dEducation Journal. 2010;38(1):33±4 1.
43. GoffJ,Hall J,Sylva K,Smith T,Smith G,Eisenstad tN, etal. Evaluati onofChildren 'sCentres in England (ECCE)Strand3:Delivery ofFamily Services byChildren 'sCentres. Oxford:Departm entfor Educatio n,2013.
Community -basedclusterRCTofthe implementa tionofan injury preventi onbriefing PLOS ONE|https://doi.or g/10.1371/journal.po ne.0172584March 24,2017 23/ 23 Copyright ofPLoS ONEisthe property ofPublic Library ofScience anditscontent maynot be copied oremailed tomultiple sitesorposted toalistserv without thecopyright holder's express writtenpermission. However,usersmayprint, download, oremail articles for individual use.