PaperType:SystematicReviewTitle:SystematicReviewoftheUseofBehaviourChangeTechniquesinPhysicalActivityeHealthInterventionsforPeoplewithCardiovascularDiseaseAuthornames:OrlaithDuff1,2,3,DeirdreWalsh1,2,3,BrónaFurlong1,2,NoelE.O’Connor3,KieranMoran2,3andCatherineWoods1,4

Affiliations:1MedExResearchCluster,DublinCityUniversity,Dublin,Ireland.2SchoolofHealthandHumanPerformance,DublinCityUniversity,Dublin,Ireland.3InsightCentreforDataAnalytics,DublinCityUniversity,Dublin,Ireland.4HealthResearchInstitute,DepartmentofPhysicalEducationandSportsSciences,FacultyofEducationandHealthSciences,UniversityofLimerick,Ireland. Correspondingauthor:KieranMoran,Kieran.moran@dcu.ie,RoomA246(AlbertCollege),SchoolofHealthandHumanPerformance,DublinCityUniversity,Dublin9,Ireland.00353–1–7008011 AbstractBackground:Cardiovasculardisease(CVD)istheleadingcauseofprematuredeathanddisability in Europe, accounting for four million deaths per year and costing the EUeconomy almost €196 billion annually. There is strong evidence to suggest thatexercise-based secondary rehabilitation programmes can decrease the mortality riskand increase health among patients with CVD. Theory informed use of behaviourchange techniques (BCTs) is important in the design of cardiac rehabilitationprogrammesaimedatchangingcardiovascularriskfactors.Electronichealth(eHealth),is the use of information and communication technologies (ICT) for health. Thisemerging area of healthcare has the ability to enhance self-management of chronicdisease through making healthcare more accessible, affordable and available to thepublic. However, evidence-based information on the use of BCTs in eHealthinterventionsislimited,andparticularlysoforindividualslivingwithCVD.Aim:Theaimof this systematic reviewwas toassess theapplicationBCTs ineHealthinterventionsdesignedtoincreasephysicalactivity(PA)inCVDpopulations.Methods: Seven electronic databases EBSCOhost (MEDLINE, PsycINFO, AcademicSearchComplete,SPORTDiscuswithFullText,CINAHLComplete),ScopusandWebofScience (Core Collection) were searched. Two authors independently reviewedreferences using the software package Covidence. The reviewers met to resolve anydiscrepancies,withathirdindependentactingasarbitratorwhenrequired.

2

Followingthis,datawasextractedfromthepapersthatmettheinclusioncriteria.Biasassessment of the studieswas carried out using the Cochrane Collaboration’s risk ofbiastoolwithinCovidence,thiswasfollowedbyanarrativesynthesis.Results:Fromthe987studiesidentified14wereincludedinthereview.Anadditional9studieswere added following a hand search of reviewpaper references. The averagenumberofBCT’susedacrossthe23studieswas7.2(range1to19).Thetopthreemostfrequently usedBCTs included; information about health consequences (78.3%), goalsetting(behaviour)(73.9%)andself-monitoringofbehaviour(47.8%).Conclusion:ThissystematicreviewisthefirsttoinvestigatetheuseofBCTsinphysicalactivityeHealthinterventionsspecificallydesignedforpeoplewithCVD.Thisresearchwill have clear implications for healthcare, policy and researchbyoutlining theBCTsusedineHealthinterventionsforchronicillnesses,inparticularCVD.Hence,providingclearfoundationsforfurtherresearchanddevelopmentsinthearea.Systematicreviewregistration:PROSPEROCRD42016034157Keywords: Systematic review,physical activity, behaviour change techniques, eHealthintervention,cardiovasculardisease

3

IntroductionCardiovasculardisease (CVD) is the leading causeofmortalityworldwide, accountingfor30%ofglobaldeathand48%ofdeathsinEurope[24].Cardiacrehabilitation(CR)isused to reduce the impact of CVD and to promote healthy behaviours and activelifestyles for thosewith CVD [25]. It has been shown to improve physical health anddecreasesubsequentmorbidityandmortalityratesinCVDpopulations[26].Themainmodalityofcardiacrehabilitationisexercise.Twosystematicreviewsofexercise-basedCR, which included 48 randomised controlled trials, showed a 20% reduction in all-causemortalityanda27%reductionincardiacmortalityattwotofiveyears[27][28].Theefficacyofstandardcardiacrehabilitationhasbeenextensivelyreviewed.Intermsofmortalityratesasystematicreviewandmeta-analysisof25randomisedcontrolledtrials(n=6111myocardialinfarctionpatients)showedthatthosewhoattendedCRhadalower-riskofall-causemortalitythannon-attendees(oddsratio0.74(0.58to0.95))[29].Intermsofhospitaladmissions,aCochranereviewof33RCTs(n=4740patientswithheartfailure,CRreducedtheriskofoverallhospitalization(relativerisk0.75(0.62to0.92),ARR7.1%,NNT15)andhospitalizationforheartfailure(relativerisk0.61(0.46to0.80),ARR5.8%,NNT18)[30].AUSobservationalstudy(n=635coronaryheartdiseasepatients)reportedimprovementsindepression,anxietyandhospitalscoresafterCR[31].Cardiacrehabilitationhasalsobeenfoundtoimprovepsychologicalwellbeingandimprovementinqualityoflife.Oneofthemostsignificantbenefitsofcardiacrehabilitationexercisetrainingtoparticipantsistheimprovementinaerobiccapacityandcardio-respiratoryfitness[32].Even though CR has been shown to be effective, adherence to these programmes isgenerally suboptimal. Participation rates in CR are documented at less than 50%worldwide [33]. Results from a Cochrane systematic review revealed that commonbarriers to adherence to CR programmes included accessibility and parking at localhospitals,adislikeofgroupenvironmentsandworkordomesticcommitments[26].In2012, aHEART journal editorial concluded that CR shouldnot only focus on content,such as coronary heart disease (CHD) risk factor modification and medicationadherence but should also focus on the delivery mechanisms, offering a range ofdifferent delivery methods for people according to their preferences and needs,potentiallyaddressingtheissueoflowlevelsofparticipation[34].ThedeliveryofCRtodatehaslargelybeencentre-based,eitherinhospitalsorcommunitycentres.However,inmorerecenttimestherehasbeenashifttowardamorehome-basedmodelofcare.AsystematicreviewbyDalalandcolleagues[26]foundthatbothhomeandcenterbasedformsofCRareequallyeffectiveinimprovingclinicalandhealthrelatedqualityoflifeoutcomes in patientswith cardiovascular disease, suggesting the further provision ofadditionalevidence-basedhomeCRprogrammes.ACochranereviewfoundthathome-basedinterventionsmaybesuperiorintermsofadherencetoexercise,especiallyinthelongterm[35].Thiswouldensurethatpatientsaregiventhechoiceofparticipatingina

4

more traditional supervised center-based programme or a home-based programme,basedontheirpersonalpreference.Theemergingareaofelectronichealth(eHealth),definedastheuseofinformationandcommunication technologies (ICT) forhealth [36]mayprovide this alternativehome-based delivery method. Interventions that encompass ICT (e.g. internet- and mobilebased communications, wearable monitors etc.) enable the efficient delivery ofeducational resources, individually tailored health and wellness programs as well astime-unlimitedfeedback,coachingandsupport[37].Technologysolutionsforphysicalactivity uptake and monitoring are being undertaken as a new mode of facilitatingbehaviour change andmay impact the current delivery of cardiac rehabilitation [38].Tele-rehabilitation solutions refer to the use of Information and CommunicationTechnologies(ICT) toproviderehabilitationservices topeople.Literature in thisareafor cardiac patients indicates that such interventions are feasible and effectivewhencomparedtoconventionalcentre-basedCR[39].

eHealth interventions have been showing promising results in cardiac rehabilitation,supportingbehaviourchange,clinicalimprovementandimprovedsocialfunctioning.In2013, Beatty and Colleagues [40] conducted a review of mobile interventions forcardiacrehabiitation,identifyingonly3studiesforinclusion.Morerecentlytheinterestin e- andmHealth has risendramaticaally, indicating the increased focus in this fieldover recent years. Buys and colleagues [38] investigated the interest among cardiacpatients in technologyenabled cardiovascular rehabilitation.Of the298patient (77%male;meanage61.7±14.5years)questionnaires includedintheanalysis,97%hadamobilephoneand91%usedtheinternet.Physicalactivitymonitoringwasreportedby12%of the respondents.Overall cardiacpatients showedhigh interest inCR supportthroughtheinternet(77%)andmobilephones(68%).ThesefindingssuggestthatCVDpatients show an interest in technology enabledhome-basedCR, potentially allowingexercise based rehabilitation programmes be more effective by making them moreaccessible,personalisedandmoreinteractivewithpatients.

BCTsareintegralinthedesignofcomplexhealthserviceinterventions,suchascardiacrehabilitation.ABCTisdefinedas“anobservable,replicableandirreduciblecomponentof an intervention designed to alter or redirect causal processes that regulatebehaviour;thatis,atechniqueisproposedtobean‘activeingredient’”[41].TheMedicalResearch Council (MRC) guidelines recommend the application of behaviour changetheory within complex health service interventions to allow for a theoreticalunderstanding of behaviour change [42]. The National Institute of Health and CareExcellence[43]guidelinesonindividual-levelbehaviourchangeinterventionsaimedatchanging health-damaging behaviours such as unhealthy diet, physical inactivity,excessive alcohol consumption, unsafe sex and smoking, recommend the use ofevidence-based BCTs, which have been proven to be effective at changing behavior,such as goals andplanning, feedback andmonitoring and social support.Despite this

5

guidance, few interventions pay close attention to the behaviour change theory andtechniques used to design their interventions. In particular, the poor description ofinterventionsinresearchprotocolsandpublishedreportspresentsabarrierforfuturedesign of complex interventions [44], as it is difficult to identify the active, effectivecomponents of the intervention [41]. The proliferation of eHealth interventionsrequires the coding of such interventions to facilitate future research to compareaccurately across interventions. With that in mind, this systematic review aims toidentify the key behaviour change techniques applied in eHealth physical activityinterventionsforadultswithcardiovasculardisease.

MethodsThis systematic review is reported in line with the Preferred Reporting Items forSystematic Reviews andMeta-analyses (PRISMA) guidance. The inclusion criteria forstudies were as follows; human randomised and quasi-randomised controlled trials,published and unpublished, of physical activity eHealth interventions for adults (³18yearsold)clinicallydiagnosedwithcardiovasculardisease.Studieswereincludedifthemain intervention component was delivered via a computer, smartphone, tablet orphone(e.g.mobilephoneApp,emails,textmessages,phonecalls)withtheprimaryorsecondary aim of increasing the physical activity level of the user. The interventionscouldbedeliveredtogroupsorindividuals.Theinclusioncriteriawaskeptquitebroadinordertoidentifyasmanystudiesaspossiblewhichhadphysicalactivityasaprimaryor secondary outcome, as well as studies which had PA as a component of theintervention.TheBehaviourChangeTaxonomyv1wasusedtoidentifythespecificBCT’susedwithintheincludedstudies[41].TworesearcherscodedfortheBCTsusingthetaxonomy.OutcomemeasuresA description of theBCTs and their frequency of use in the 23 eHealth interventionsreviewedwereclassifiedusingMichie’staxonomy.Duetotheheterogeneousnatureofthe studies,differing inphysicalactivityoutcomemeasuresand time-pointswewereunabletocarryoutameta-analysisexaminingtheeffectivenessoftheBCTsinrelationtothephysicalactivityoutcomes.SearchmethodsfortheidentificationofstudiesSevenelectronicdatabasesweresearched,includingMEDLINE(viaEbscoHost,2000to2016), PsycINFO (via EbscoHost, 2000 to 2016), Academic Search Complete (viaEbscoHost, 2000 to 2016), SPORTDiscus (via EbscoHost, 2000 to 2016), CINAHLComplete (via EbscoHost, 2000 to 2016), Scopus (2000 to 2016) andWebof Science(CoreCollection)(2000to2016).

6

ThesearchwasrestrictedtostudiespublishedinEnglishbetween2000and2016.Thesearch strategy used keywords relating to physical activity, eHealth interventions,cardiovasculardiseaseandadults,aswellasappropriatesynonyms.Booleanoperatorswereusedtoexpand,excludeorjoinkeywordsinthesearch,usingtheterms“AND”and“OR”.Inalldatabases,thesearcheswerelimitedtothefieldsofabstractandtitleonly.ThesearchstrategyforalldatabasesisillustratedintheAdditionalFile1.

SelectionofstudiesFigure 1 shows the PRISMA flow diagram of reviewed and included studies. Oneresearcherconductedthedatabasesearch.Allarticlesidentifiedfollowingthedatabasesearch were then uploaded to the online systematic review software package“Covidence”. Firstly, a title and abstract review of all studies was completedindependently by two authors. Any disagreements were discussed until a consensuswasreachedorathirdreviewerhelpedtoresolvethediscrepancy.Arecordwaskeptof all the articles excluded and the reason for exclusion via Covidence. Secondly, allarticlesthatmettheinclusioncriteriawentthroughafulltextscreeningprocessbythetwo authors independently. Again, any disagreements between the authors on theeligibilityofthestudieswerereviewedbyathirdauthor.Additionalstudieswerealsoidentifiedforinclusionbyreviewingthereferencelistsofreviewpapersthroughahandsearch.

7

Figure1.PRISMAflowdiagramofreviewedandincludedstudiesDataextractionDatafromthestudieswereextractedindependentlybytworeviewauthorsusingadataextraction template. Data extracted from the articles included study title, authors,country,year,patientgroup(samplesize), inclusioncriteria, studydesign, technologyinvolvement, assessment, intervention details, outcomes, theory involved, BCTsidentified and results. No blinding to study author, institution or journal occurredduringthestudyscreeningprocess.Ifmultiplepublicationsofthesamestudywereidentified,theteamwouldtrytoextractandcombinealltheavailabledataandwheretherewasdoubt,theoriginalpublicationwouldbegivenpriority. Ifdataseemedtobemissingfromastudy,wetriedtoobtainthis through correspondence with the study authors. The review team resolved anydisagreementsregardingstudyeligibilitythroughgroupdiscussion.AssessmentofriskbiasTwo reviewers assessed each study for risk of bias (high, low or unclear) using theCochraneriskofbias tool [45].A thirdreviewauthoractedasarbitrator ifnecessary.

8

The results of the risk of bias assessmentwere then exported toRevMan to create avisualrepresentationofthepublicationbias(seeFigure2).

Figure 2: Risk of bias graph: review authors' judgements about each risk of bias item presented aspercentagesacrossallincludedstudies.AssessingforheterogeneityDiversityacrossthestudieswasassessedqualitativelyintermsofeHealthintervention,patientcharacteristicsandoutcomemeasures.DatasynthesisFollowingtheextractionofdatafromthestudies,carefulconsiderationwasgiventotheappropriatenessofconductingameta-analysis.Asthestudiesweretooheterogeneoustocombinestatistically,thedataweresynthesisedqualitatively.BehaviourChangeTechniques(BCTs)Togainanunderstandingofthetypesofbehaviourchangetechniquesusedinphysicalactivity eHealth interventions in this patient population, two authors screened theincludedarticles’andcodedthebehaviourchangetechniques(BCTs)usedineachstudyusingMichie’sBCTtaxonomy[41]. ResultsThesearchcriteriareturned1391articlesthroughdatabasessearching.Atotalof404duplicateswereremoved,leaving987articlestoscreen.Thearticlestitleandabstractswerethenscreenedbytworeviewers,resultingin891recordsexcludedfornotmeetingthe inclusioncriteria. Theauthors reviewed the full textof96studies, identifying14studiesforinclusioninthisreview.Fromahandsearchofreviewpapersreferencesanadditional58studieswereidentifiedaspotentiallyeligible.Followingafulltextreview

9

ofthesepapers,9studieswereincludedinthereview.Therefore,atotalof23articleswereincludedinthequalitativesynthesis.StudycharacteristicsTable1providesanoverviewoftheincludedstudiesandthephysicalactivityresults.Ofthe 23 studies included, 14 comprised an internet/web-based programme and/orsmartphone intervention [2-3, 6, 8, 10-11, 14, 16-19, 21-23], 3 were telephoneinterventions[1,12-13],2usedatelehealthdevice[4][5],and2consistedofaformoftelemonitoring [15] [20]. Single studies consisting of videoconferencing [9] and ofvirtualrealitywraparoundscreens[7]werealsofound.Ofthe20studieswithacontrolgroup, 17 involved ‘usual care’ as the control. Usual care predominately pertained toreceiving standard cardiac rehabilitation services [1,4-11, 13-15, 17, 19,21-23]. Eightstudies were carried out in Europe [2, 3, 10-11, 13, 16, 18, 20], while seven of thestudieswereconductedinNorth/SouthAmerica[1,4-5,14,18,21,23].Threestudiesapiece were conducted in Australia [6, 12, 22] and New Zealand [8-9, 17] and twostudieswereconductedinAsia[6,15].Themajorityofparticipantswererecruitedfromhospitals/medicalcentres[1-8,10,12-14, 16, 18-23]. One study recruited participants from a general practitioner (GP)coronaryheartdisease(CHD)registry[16],whileanotherrecruitedfromaCRreferrallist[12].Tomitaandcolleagues[21]recruitedparticipantsfromthreehospitalsandtwohealth insurance companies. One study recruited participants from primary andcommunity health services [22]. Outcomes were assessed from 3 weeks [5] to 16months[13],withtheaverageend-pointacrossthe23studiesat4.5months.-InsertTable1approximatelyhere-

10

BehaviouralchangetechniquesOnly2outof the23studiesexplicitlymentioned theBCTsapplied [8] [10].Fromtheother studies, two reviewers coded the BCTs from the programdescription. Table 2outlines the number of BCTs used in each study as well a comprehensive list of thetechniquesused.TheaveragenumberofBCTsemployedintheincludedstudieswas7.2(Range1-14).ThetopthreemostfrequentlyusedBCTswereidentifiedasinformationabout health consequences (78.3%), goal setting (behaviour) (73.9%) and self-monitoring of behaviour (47.8%) (See Table 2). The Text4Heart study conducted byDaleandcolleagues[8]employedthemostBCT’soutofallthearticles,using14.Thesewere goal setting (behaviour), problem solving, review outcome goals, feedback onbehaviour, self-monitoring of behaviour, social support (unspecified), instruction onhowtoperformthebehaviour,informationabouthealthconsequences,demonstrationof thebehaviour, social comparison, prompts/cues, graded tasks, credible source andreduce negative emotions. A study by Barnason and colleagues [5] used the leastamount of BCTs of the 23 studies included in the review, employing just one BCT,gradedtasks.The most common BCT group used in the 23 included studies was feedback andmonitoring,while the secondmost common groupwas goals and planning. Thiswasfollowed by social support. Four groups did not appear in any of the 23 includedstudies;identity,scheduledconsequences,self-beliefandcovertlearning.-InsertTable2approximatelyhere-

11

Table3outlinesthefrequencyofuseoftheBCTsacrossthe23studies,theBCTtaxonomygroupandanexampleofhowaBCTwasincorporatedintoastudy.OnlytwoBCTswereusedinover70%ofthestudies,thesewere5.1informationabouthealthconsequences(78.3%)and1.1goalsetting(behaviour)(73.9%).Afurther4BCTswereusedinover40%ofthestudies,theseinclude;2.2feedbackonbehaviour(43.5%),2.3self-monitoringofbehaviour(47.8%),3.2socialsupport(practical)(47.8%)and4.1instructiononhowtoperformthebehaviour(43.5%).SeveralBCTsincluding,10.3non-specificreward,12.1restructuringthephysicalenvironment,12.5addingobjectstotheenvironment,11.1pharmacologicalsupport,6.1demonstrationofthebehaviour,6.2socialcomparison,1.7reviewoutcomegoals,10.4socialrewardand1.8behaviouralcontractwereonlyusedinonestudy(SeeTable3formoredetails).-InsertTable3approximatelyhere-

12

BehaviourChangeTechniquesLinkedtoImprovedPhysicalActivityOutcomesEightofthe15interventionsthathadPAasanoutcomemeasurereportedstatisticallysignificantimprovementsinphysicalactivitybetweentheexperimentalandcontrolgroups.Goalsetting(behaviour)andinformationabouthealthconsequenceswerethemostfrequentlyusedBCTsacrosstheeightstudies(n=6each).Thiswasfollowedbyfeedbackonbehaviourandinstructiononhowtoperformthebehaviour,whichwereincorporatedin5studieseach.ThefollowingBCT’swerealsoincludedintheinterventionswhichhadanimprovedPAoutcomeatthefinalendpoint;self-monitoringofbehaviour,socialsupport(practical),socialsupport(unspecified),crediblesource,problemsolving,reviewbehaviourgoals,socialsupport(emotional),prompts/cues,gradedtasks,reducenegativeemotions,actionplanning,self-monitoringofoutcomesofbehaviour,biofeedback,feedbackonoutcome(s)ofbehaviour,socialrewardandpharmacologicalsupport(Seetable4).Table4:Frequencyofbehaviouralchangetechniques(BCTs)usedinstudieswithimprovedPAoutcomeBCTlabel

Totalnumberofstudiesn=8

N(%)

1.1Goalsetting(behaviour) 6(75)

5.1Informationabouthealthconsequences 6(75)

2.2Feedbackonbehaviour 5(62.5)4.1Instructiononhowtoperformthebehaviour 5(62.5)2.3Self-monitoringofbehaviour 4(50)3.2Socialsupport(practical) 4(50)3.1Socialsupport(unspecified) 3(37.5)9.1Crediblesource 3(37.5)

1.2Problemsolving 2(25)1.5Reviewbehaviourgoals 2(25)

3.3Socialsupport(emotional) 2(25)

7.1Prompts/cues 2(25)

8.7Gradedtasks 2(25)

11.2Reducenegativeemotions 2(25)

1.4Actionplanning 1(12.5)

13

2.4Self-monitoringofoutcomesofbehaviour 1(12.5)

2.6Biofeedback 1(12.5)

2.7Feedbackonoutcome(s)ofbehaviour 1(12.5)

10.4Socialreward 1(12.5)

11.1Pharmacologicalsupport 1(12.5)

1.3Goalsetting(outcome) 0(0)

1.7Reviewoutcomegoals 0(0)

1.8Behaviouralcontract 0(0)

2.1Monitoringofbehaviourbyotherswithoutfeedback

0(0)

2.5Monitoringofoutcomesofbehaviourwithoutfeedback

0(0)

6.1Demonstrationofthebehaviour 0(0)

6.2Socialcomparison 0(0)

10.3Non-specificreward 0(0)

12.1Restructuringthephysicalenvironment 0(0)

12.5Addingobjectstotheenvironment 0(0)

ItisworthnotingthatthoseinterventionsthatdidnotdiddemonstrateasignificantincreaseinPA(n=5)wereonparwiththelevelachievedinstandardCR,asnosignificantdifferencesbetweenthecontrolandexperimentalgroupswerefound.ThisisanimportantfindingasithighlightsthefactthattheeHealthinterventionswereonparwithorweresignificantlybetteratimprovingPAlevelsofcardiacpatientswhencomparedtostandardcardiacservices.ThisemphasizesthepotentialofeHealthinterventionsinacardiacrehabilitationsetting.TofurtherexaminetheefficacyoftheindividualBCTstheinterventionsweregroupedinto fourgroupsdependingonwhetherphysicalactivitywasmeasuredobjectivelyorsubjectively and whether there was a difference between experimental and controlgroups.Oncethe interventionsweregroupedwesoughttoexamine if therewereanycommonBCTsusedacrossthestudies(Seetable5).Thistaskallowedustoexamineifthere were any similarities between the interventions in terms of the BCTs they

14

employed.Objective and self-report studieswith nodifference between experimentalandcontrolgroupsweretheonlygroupswithsimilarities intheBCTstheyemployed.Socialsupport(practical)andinformationandhealthconsequenceswereemployedinallself-reportstudieswheretherewasnoPAdifferencebetweentheexperimentalandcontrolgroups.Goalsetting(behaviour)andfeedbackonbehaviourwereemployedinall PA objectively measured intervention where no significant difference was foundbetweengroupsatthefinalendpoint.However,therewerenosimilaritiesintheBCTsused across all the effective interventions, regardless of whether PA was measuredobjectively or subjectively. Furthermore, the average number of BCTs used acrosssignificant interventionsdidnotdiffer, as studies that increasedPAversus those thatdidnotincreasePAemployedonaverage7BCTs. -InsertTable5approximatelyhere-

15

DiscussionSummaryThis systematic review consisted of 23 studies reviewing the use of BCTs in physicalactivityeHealthinterventionsforadultswithcardiovasculardisease.Toourknowledge,this is the first review that aimed to identify the use of Michie’s behaviour changetaxonomyinphysicalactivityeHealthinterventionstudiesamongthispopulation.Thefindings of the review indicate that an average of 7.2BCTswere employed in the 23studies. Information about health consequences was the most frequently usedtechnique,with78.3%of studies incorporating this technique into their intervention.Thiswasfollowedcloselybygoalsetting(behaviour),whichwasusedin73.9%ofthestudies,withself-monitoringofbehaviouremployedin47.8%ofthestudies.Although Michie’s behaviour change technique taxonomy is made up of 93 differenttechniques, themaximumamount of techniquesused in a single interventionwas14[8]. These were goal setting (behaviour), problem solving, review outcome goals,feedback on behaviour, self-monitoring of behaviour, social support (unspecified),instructiononhowtoperformthebehaviour, informationabouthealthconsequences,demonstration of the behaviour, social comparison, prompts/cues, graded tasks,credible source and reduce negative emotions. The minimum number of techniquesusedinastudywasone;gradedtasks[5].Afailingofthestudiesincludedinthisreviewwasthepoordescriptionoftheinterventioncomponents.Onlytwopapersinthereviewspecifically mentioned the behaviour change techniques incorporated in theirinterventions [8] [10]. However, even though the paper by Devi and colleagues [10]listed theBCT’s used, it failed to link theBCT’s used to the intervention functions orcomponents. In the studybyDale [8] the researchersprovidedonly examples of textmessages linked toBCTs.Neither study gave a full account of theBCTs used in theirstudies andhow thesewere linked to the intervention components.This finding is inlinewithpreviousresearch,wherereviewsofnearly1,000behaviourchangeoutcomestudiesfoundthatinterventionswerefullyandaccuratelyweredescribedinonly5%to30%ofexperimentalstudies[46][47][48][49].Overallthislackofrobustanddetailedinformation on the intervention functions provide a significant barrier to betterunderstandingtheeffectsandmechanismsofbehaviourchangeinterventions,toinformthedevelopmentofmoreeffectiveinterventionsinthefuture[39].

Anotherkeyissuerelatingtothepoordescriptionofbehaviourchangeinterventionsisthe inconsistent use of terminology. This variation in terminology used makes thecoding of the techniques used evenmore difficult when reviewing behaviour changeinterventions.Forexample,socialsupport(unspecified)wascodedforin41.67%ofthestudiesincludedinthereviewbythereviewers.Terminologyvariedacrossthestudieswhere social supportwas coded, for example, one study used a social reinforcementnetwork[3],anotherincorporatedmentorsintotheir intervention[22],whileanotherstudyinvolvedtutorialsintheirintervention[19].Thereviewerscodedtheseexamplesassocialsupport(unspecified)however,thisBCTwasnotspecificallymentionedinany

16

ofthestudies.Therefore,thereisaneedtohaveconsistentterminologyandsufficientinformation on intervention components to allow for the replication of interventionsthat have been found to be effective. The lack of such information appears to beparticularlyproblematic inbehavioural interventions rather than forpharmacologicalones [44]. In a workshop 26 multi-disciplinary researchers were presented withbehaviouralorpharmacological interventionprotocolsandwereasked if theprotocolprovided sufficient information so that the study could be replicated in a practicesetting.The researcherswere less confident that they could replicate thebehaviouralinterventionscomparedtothepharmacologicalinterventions(t=6.45,p<0.0001)andconcluded they would need more information to replication the behaviouralinterventions(U=35.5,p=0.022)[50].

This reviewprovides new and important information regarding the use of behaviourchange techniques in eHealth physical activity for adults with CVD, highlighting thefrequent use of the following BCT’s; information about health consequences, goalsetting(behaviour),self-monitoringofbehaviour.However,itisclearthatmorerobustand comprehensive interventions are needed, which systematically and coherentlydetailthebehaviourchangetechniquesusedintheinterventions.Identifyingtheactiveingredientsoftheinterventionswillenableresearcherstoexaminetheeffectivenessofthese key intervention components, ensuring that the most effective BCTs are usedregardingeHealthphysicalactivityinterventionsforadultswithcardiovasculardisease.StrengthsandLimitationsAmajorstrengthofthisreviewwastheauthorsattempttoidentifyallrelevantstudiesby using a comprehensive search strategy andmultiple databases. The authors’ alsohand searched review paper references to identify any additional studieswhichmayhavebeen relevant to the review.All articles identified following thedatabase searchwerethenuploadedtotheonlinesystematicreviewsoftwarepackage“Covidence”.Thisallowed for a systematic and comprehensive approach to screening the articles andcoding the reasons for exclusion.This software also enabled the screening for riskofbiasinasimpleandefficientway.Fromthis,avisualrepresentationofthepublicationbiaswasproducedusingRevMan.A limitation of this reviewwas thewide variability among the studies included,withstudydesignsrangingfromrandomisedcontrolledtrials,tofeasibilitystudiesandpilottrials.However,itwasnecessarytoincludeallstudiesandnotjustRCTstoidentifyasmany physical activity eHealth interventions as possible. There was also a lack ofconsistencyinthemeasurementofphysicalactivityacrossthestudies,fromsubjectivetoobjectiveassessments.Thefollow-updurationalsovariedsignificantlyfrom3weeksto16months.Thismeantitwasimpossibletopooltheresultsinameta-analysis.

17

Many studiesmeasured the physical fitness of their participants, as opposed to theirphysical activity levels.Althoughall the interventionshadaphysical activity/exercisecomponent to their eHealth intervention, some studies did not directly measure thephysicalactivitylevelofparticipants.Wecanthereforeonlyinferfromthestudiesthatby increasingphysical activity behaviour that the physical fitness outcome improved.Thisinferenceofacausalrelationshipbetweenphysicalactivityandphysicalfitnessisalimitation to these studies. Another limitation is the variety of methods used tomeasurephysicalactivity,meaningthatcomparisonbetweenstudiesischallengingandthereforedeterminingtheimpactofspecificBCTsisimpossible.ImplicationsforresearchandpracticeThissystematicreviewhighlightstheneedformorerobustandcomprehensiveeHealthphysical activity interventions for adults with CVD. While the most frequently usedBCT’sare identified, it isworthnoting that themajorityofstudiesdidnotspecificallydetail the active ingredients of their interventions. Further work is also needed todeterminewhat is themostappropriatemeasurementofphysical activityamong thispopulationsothatinterventionsusethebestsubjectiveand/orobjectivemeasurementsensuringcomparisonscanbeeasilydrawnacross studies.The reviewalsohighlightstheimportanceofidentifyingthebehaviourchangetechniquesusedwithinastudyandtheir link to the interventioncomponents inorder tounderstand the ingredients thatbringaboutthedesiredbehaviourchange.Itisonlybyidentifyingthesemechanismsofchangethatwecanunderstandwhyaninterventionwasfoundtobeeffectiveornot.Authors’contributionsOD ran the keyword search in the chosen databases and reviewed all articles forinclusionandexclusion.ODdraftedinlargepartthefirstversionofthemanuscript.DWwas thesecondreviewerwhoreviewed thearticles for inclusionandexclusion in thereview.DWandOD independently extracteddata from the final papers for inclusion.CWwas the third reviewer if anydiscrepanciesoccurredbetweenODandDW in thereviewanddataextractionprocesses.CW,DWandBFrevisedandprovided feedbackon the drafts on the manuscript. KW and NOC also provided feedback on themanuscript.Allauthorshavereadandapprovedthefinalversionofthemanuscript.FundingThisresearchhasbeenfundedbyScienceFoundationIreland(SFI)undergrantnumberSFI/12/RC/2289andourindustrypartnerAcquis-bibasedinLosGatos,California. ConflictsofinterestNonedeclared.AbbreviationsBCTs:Behaviourchangetechniques

18

CVD:CardiovasculardiseaseCR:CardiacrehabilitationCHD:CoronaryheartdiseaseeHealth:ElectronichealthICT:InformationandcommunicationtechnologiesPA:PhysicalactivitySFI:ScienceFoundationIreland

19

BibliographyReferencesofreviewarticles

1. Ades,P.A.,Pashkow,F.J.,Fletcher,G.,Pina,I.L.,Zohman,L.R.,&Nestor,J.R.(2000).Acontrolledtrialofcardiacrehabilitationinthehomesettingusingelectrocardiographicandvoicetranstelephonicmonitoringhttp://dx.doi.org/10.1016/S0002-8703(00)90100-5

2. Ammenwerth,E.,Woess,S.,Baumgartner,C.,Fetz,B.,vanderHeidt,A.,Kastner,P.,etal.(2015).Evaluationofanintegratedtelemonitoringsurveillancesysteminpatientswithcoronaryheartdisease.MethodsofInformationinMedicine,54(5),388-397.https://doi.org/10.3414/ME15-02-0002

3. Antypas,K.,&Wangberg,S.C.(2014).Aninternet-andmobile-basedtailoredinterventiontoenhancemaintenanceofphysicalactivityaftercardiacrehabilitation:Short-termresultsofarandomizedcontrolledtrial.JournalofMedicalInternetResearch,16(3),78-95.https://doi.org/10.2196/jmir.3132

4. Artinian,N.T.,Harden,J.K.,Kronenberg,M.W.,VanderWal,J.,S.,Daher,E.,Stephens,Q.,etal.(2003).Pilotstudyofaweb-basedcompliancemonitoringdeviceforpatientswithcongestiveheartfailure.Heart&Lung:TheJournalofCriticalCare,32(4),226-233.Retrievedfromhttp://search.ebscohost.com/login.aspx?direct=true&db=cmedm&AN=12891162&site=ehost-livehttps://doi.org/10.1016/S0147-9563(03)00026-8

5. Barnason,S.,Zimmerman,L.,Nieveen,J.,Schulz,P.,Miller,C.,Hertzog,M.,etal.(2009).Influenceofasymptommanagementtelehealthinterventiononolderadults'earlyrecoveryoutcomesfollowingcoronaryarterybypasssurgery(CABS).Heart&Lung:TheJournalofCriticalCare,38(5),364-376.https://doi.org/10.1016/j.hrtlng.2009.01.005

6. Chow,C.K.,Redfern,J.,Hillis,G.S.,Thakkar,J.,Santo,K.,Hackett,M.L.,etal.(2015).Effectoflifestyle-focusedtextmessagingonriskfactormodificationinpatientswithcoronaryheartdisease:Arandomizedclinicaltrial.JAMA-JournaloftheAmericanMedicalAssociation,314(12),1255-1263.RetrievedfromSCOPUSdatabase.https://doi.org/10.1001/jama.2015.10945PMid:26393848

7. Chuang,T.,Sung,W.,Chang,H.,&Wang,R.(2006).Effectofavirtualreality-enhancedexerciseprotocolaftercoronaryarterybypassgrafting.PhysicalTherapy,86(10),1369-13779p.https://doi.org/10.2522/ptj.20050335

8. Dale,L.P.,Whittaker,R.,Jiang,Y.,Stewart,R.,Rolleston,A.,&Maddison,R.(2015).Textmessageandinternetsupportforcoronaryheartdiseaseself-management:ResultsfromtheText4Heartrandomizedcontrolledtrial.JournalofMedicalInternetResearch,17(10),e237.https://doi.org/10.2196/jmir.4944

9. Dalleck,L.C.,Schmidt,L.K.,&Lueker,R.(2011).Cardiacrehabilitationoutcomesinaconventionalversustelemedicine-basedprogramme.JournalofTelemedicineandTelecare,17(5),217-221.https://doi.org/10.1258/jtt.2010.100407

20

10. Devi,R.,Powell,J.,&Singh,S.(2014).Aweb-basedprogramimprovesphysicalactivityoutcomesinaprimarycareanginapopulation:Randomizedcontrolledtrial.JMedInternetRes,16(9),e186.https://doi.org/10.2196/jmir.3340

11. Frederix,I.,Hansen,D.,Coninx,K.,Vandervoort,P.,Vandijck,D.,Hens,N.,etal.(2015).Medium-termeffectivenessofacomprehensiveinternet-basedandpatient-specifictelerehabilitationprogramwithtextmessagingsupportforcardiacpatients:Randomizedcontrolledtrial.JournalofMedicalInternetResearch,17(7),e185.https://doi.org/10.2196/jmir.4799

12. Furber,S.,Butler,L.,Phongsavan,P.,Mark,A.,&Bauman,A.(2010).Randomisedcontrolledtrialofapedometer-basedtelephoneinterventiontoincreasephysicalactivityamongcardiacpatientsnotattendingcardiacrehabilitation.PatientEducationandCounseling,80(2),212-218.http://dx.doi.org/10.1016/j.pec.2009.11.012

13. Hanssen,T.A.,Nordrehaug,J.E.,Eide,G.E.,&Hanestad,B.R.(2007).Improvingoutcomesaftermyocardialinfarction:arandomizedcontrolledtrialevaluatingeffectsofatelephonefollow-upintervention.EuropeanJournalofCardiovascularPrevention&Rehabilitation,14(3),429-437.https://doi.org/10.1097/HJR.0b013e32801da123

14. Lear,S.A.,Singer,J.,Banner-Lukaris,D.,Horvat,D.,Park,J.E.,Bates,J.,etal.(2014).Randomizedtrialofavirtualcardiacrehabilitationprogramdeliveredatadistanceviatheinternet.Circulation:CardiovascularQualityandOutcomes,7(6),952-959.https://doi.org/10.1161/CIRCOUTCOMES.114.001230

15. Lee,Y.,Hur,S.,Sohn,J.,Lee,H.,Park,N.,Cho,Y.,etal.(2013).Impactofhome-basedexercisetrainingwithwirelessmonitoringonpatientswithacutecoronarysyndromeundergoingpercutaneouscoronaryintervention.JournalofKoreanMedicalScience,28(4),564-568.https://doi.org/10.3346/jkms.2013.28.4.564

16. Lindsay,S.,Smith,S.,Bellaby,P.,&Baker,R.(2009).Thehealthimpactofanonlineheartdiseasesupportgroup:Acomparisonofmoderatedversusunmoderatedsupport.HealthEducationResearch,24(4),646-654.https://doi.org/10.1093/her/cyp001

17. Maddison,R.,Pfaeffli,L.,Whittaker,R.,Stewart,R.,Kerr,A.,Jiang,Y.,etal.(2015).Amobilephoneinterventionincreasesphysicalactivityinpeoplewithcardiovasculardisease:ResultsfromtheHEARTrandomizedcontrolledtrial.EuropeanJournalofPreventiveCardiology,22(6),701-709.RetrievedfromSCOPUSdatabase.https://doi.org/10.1177/2047487314535076PMid:24817694

18. Piotrowicz,E.,Korzeniowska-Kubacka,I.,Chrapowicka,A.,Wolszakiewicz,J.,Dobraszkiewicz-Wasilewska,B.,Batogowski,M.,etal.(2014).Feasibilityofhome-basedcardiactelerehabilitation:ResultsofTeleInterMedstudy.CardiologyJournal,21(5),539-546.https://doi.org/10.5603/CJ.a2014.0005

19. Reid,R.D.,Morrin,L.I.,Beaton,L.J.,Papadakis,S.,Kocourek,J.,McDonnell,L.,etal.(2012).Randomizedtrialofaninternet-basedcomputer-tailoredexpert

21

systemforphysicalactivityinpatientswithheartdisease.EuropeanJournalofPreventiveCardiology,19(6),1357-1364.RetrievedfromSCOPUSdatabase.https://doi.org/10.1177/1741826711422988PMid:21903744

20. Scalvini,S.,Zanelli,E.,Comini,L.,Tomba,M.D.,Troise,G.,&Giordano,A.(2009).Home-basedexerciserehabilitationwithtelemedicinefollowingcardiacsurgery.JournalofTelemedicineandTelecare,15(6),297-301.https://doi.org/10.1258/jtt.2009.090208

21. Tomita,M.R.,Tsai,B.,Fisher,N.M.,Kumar,N.A.,Wilding,G.,Stanton,K.,etal.(2008).Effectsofmultidisciplinaryinternet-basedprogramonmanagementofheartfailure.JournalofMultidisciplinaryHealthcare,2,13-21.Retrievedfromhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC2875686/https://doi.org/10.2147/JMDH.S4355

22. Varnfield,M.,Karunanithi,M.,Lee,C.,Honeyman,E.,Arnold,D.,Ding,H.,etal.(2014).Smartphone-basedhomecaremodelimproveduseofcardiacrehabilitationinpostmyocardialinfarctionpatients:Resultsfromarandomisedcontrolledtrial.Heart,100(22),1770-1779.https://doi.org/10.1136/heartjnl-2014-305783

23. Widmer,R.J.,Allison,T.G.,Lerman,L.O.,&Lerman,A.(2015).Digitalhealthinterventionasanadjuncttocardiacrehabilitationreducescardiovascularriskfactorsandrehospitalizations.JournalofCardiovascularTranslationalResearch,8(5),283-292.https://doi.org/10.1007/s12265-015-9629-1

AdditionalReferences

24. Allender,S.,Scarborough,P.,Peto,V.,Rayner,M.,Leal,J.,Luengo-Fernandez,R.,etal.(2008).Europeancardiovasculardiseasestatistics.

25. Balady,G.J.,Ades,P.A.,Bazzarre,T.,Comoss,P.,Limacher,M.,Pina,I.L.,etal.(2000).CorecomponentsofcardiacRehabilitation/Secondarypreventionprograms:Statementforhealthcareprofessionalsfromtheamericanheartassociationandtheamericanassociationofcardiovascularandpulmonaryrehabilitation.JournalofCardiopulmonaryRehabilitationandPrevention,20(5)Retrievedfromhttp://journals.lww.com/jcrjournal/Fulltext/2000/09000/Core_Components_of_Cardiac.76.aspxhttps://doi.org/10.1097/00008483-200009000-00076

26. Dalal,H.M.,Zawada,A.,Jolly,K.,Moxham,T.,&Taylor,R.S.(2010).Homebasedversuscentrebasedcardiacrehabilitation:Cochranesystematicreviewandmeta-analysis.BMJ(ClinicalResearchEd.),340,b5631.https://doi.org/10.1136/bmj.b5631

27. Jolliffe,J.,Rees,K.,Taylor,R.R.,Thompson,D.R.,Oldridge,N.,&Ebrahim,S.(2001).Exercise-basedrehabilitationforcoronaryheartdisease.TheCochraneLibrary,https://doi.org/10.1002/14651858.cd001800

22

28. Taylor,R.S.,Brown,A.,Ebrahim,S.,Jolliffe,J.,Noorani,H.,Rees,K.,etal.(2004).Exercise-basedrehabilitationforpatientswithcoronaryheartdisease:Systematicreviewandmeta-analysisofrandomizedcontrolledtrials.TheAmericanJournalofMedicine,116(10),682-692.https://doi.org/10.1016/j.amjmed.2004.01.009PMid:15121495

29. LawlerPR,FilionKB,EisenbergMJ.Efficacyofexercise-basedcardiacrehabilitationpost-myocardialinfarction:asystematicreviewandmeta-analysisofrandomizedcontrolledtrials.AmHeartJ2011;162:571-584.e2.

30. SagarVA,DaviesEJ,BriscoeS,CoatsAJ,DalalHM,LoughF,etal.Exercise-basedrehabilitationforheartfailure:systematicreviewandmeta-analysis.OpenHeart2015;2:e000163.

31. LavieCJ,MilaniRV.Adversepsychologicalandcoronaryriskprofilesinyoungpatientswithcoronaryarterydiseaseandbenefitsofformalcardiacrehabilitation.ArchInternMed2006;166:1878-83.

32. Menezes,A.R.,Lavie,C.J.,Milani,R.V.,Arena,R.A.,&Church,T.S.(2012).Cardiacrehabilitationandexercisetherapyintheelderly:Shouldweinvestintheaged?JournalofGeriatricCardiology :JGC,9(1),68–75.http://doi.org/10.3724/SP.J.1263.2012.00068

33. Bjarnason-Wehrens,B.,McGee,H.,Zwisler,A.,Piepoli,M.F.,Benzer,W.,Schmid,J.,etal.(2010).CardiacrehabilitationinEurope:ResultsfromtheEuropeancardiacrehabilitationinventorysurvey.EuropeanJournalofCardiovascularPrevention&Rehabilitation,17(4),410-418.https://doi.org/10.1097/HJR.0b013e328334f42d

34. Jones,R.,Furze,G.,Buckley,J.,Speck,L.,Carver,C.,Connolly,S.,etal.(2012).RAMITpresentsanoutdatedversionofcardiacrehabilitation.Heart,98(8),672-672.https://doi.org/10.1136/heartjnl-2012-301766

35. Ashworth,N.L.,Chad,K.E.,Harrison,E.L.,Reeder,B.A.,&Marshall,S.C.(2005).Homeversuscenterbasedphysicalactivityprogramsinolderadults.TheCochraneLibrary.https://doi.org/10.1002/14651858.cd004017.pub2

36. WorldHealthOrganisation.(2012).GlobalobservatoryforeHealthseries–volume5.legalframeworksforeHealth.Geneva:WHO.

37. Krebs,P.,Prochaska,J.O.,&Rossi,J.S.(2010).Ameta-analysisofcomputer-tailoredinterventionsforhealthbehaviorchange.PreventiveMedicine,51(3),214-221.https://doi.org/10.1016/j.ypmed.2010.06.004PMid:20558196PMCid:PMC2939185

38. Buys,R.,Claes,J.,Walsh,D.,Cornelis,N.,Moran,K.,Budts,W.,etal.(2016).Cardiacpatientsshowhighinterestintechnologyenabledcardiovascularrehabilitation.BMCMedicalInformaticsandDecisionMaking,16(1),95.https://doi.org/10.1186/s12911-016-0329-9PMid:27431419PMCid:PMC4949752

39. Frederix,I.,Vanhees,L.,Dendale,P.,&Goetschalckx,K.(2015).Areviewoftelerehabilitationforcardiacpatients.JournalofTelemedicineandTelecare,21(1),45-53.https://doi.org/10.1177/1357633X14562732

23

40. Beatty,A.L.,Fukuoka,Y.,&Whooley,M.A.(2013).Usingmobiletechnologyforcardiacrehabilitation:areviewandframeworkfordevelopmentandevaluation.JournaloftheAmericanHeartAssociation,2(6),e000568.

41. Michie,S.,Richardson,M.,Johnston,M.,Abraham,C.,Francis,J.,Hardeman,W.,etal.(2013).Thebehaviorchangetechniquetaxonomy(v1)of93hierarchicallyclusteredtechniques:Buildinganinternationalconsensusforthereportingofbehaviorchangeinterventions.AnnalsofBehavioralMedicine,46(1),81-95.https://doi.org/10.1007/s12160-013-9486-6PMid:23512568

42. Davis,R.,Campbell,R.,Hildon,Z.,Hobbs,L.,&Michie,S.(2015).Theoriesofbehaviourandbehaviourchangeacrossthesocialandbehaviouralsciences:Ascopingreview.HealthPsychologyReview,9(3),323-344.https://doi.org/10.1080/17437199.2014.941722PMid:25104107PMCid:PMC4566873

43. NationalInstituteforHealthandCareExcellence.(2014).Behaviourchange:Individualapproaches.NICE.

44. Michie,S.,Fixsen,D.,Grimshaw,J.M.,&Eccles,M.P.(2009).Specifyingandreportingcomplexbehaviourchangeinterventions:Theneedforascientificmethod.ImplementationScience,4(1),40.https://doi.org/10.1186/1748-5908-4-40PMid:19607700PMCid:PMC2717906

45. Higgins,J.P.,Altman,D.G.,Gotzsche,P.C.,Juni,P.,Moher,D.,Oxman,A.D.,etal.(2011).Thecochranecollaboration'stoolforassessingriskofbiasinrandomisedtrials.BMJ(ClinicalResearchEd.),343,d5928.dhttps://doi.org/10.1136/bmj.d5928

46. Dane,A.V.,&Schneider,B.H.(1998).Programintegrityinprimaryandearlysecondaryprevention:Areimplementationeffectsoutofcontrol?ClinicalPsychologyReview,18(1),23-45.https://doi.org/10.1016/S0272-7358(97)00043-3

47. Gresham,F.M.,Gansle,K.A.,&Noell,G.H.(1993).Treatmentintegrityinappliedbehavioranalysiswithchildren.JournalofAppliedBehaviorAnalysis,26(2),257-263.https://doi.org/10.1901/jaba.1993.26-257PMid:8331022PMCid:PMC1297745

48. Moncher,F.J.,&Prinz,R.J.(1991).Treatmentfidelityinoutcomestudies.ClinicalPsychologyReview,11(3),247-266.https://doi.org/10.1016/0272-7358(91)90103-2

49. Odom,S.L.,Brown,W.H.,Frey,T.,Karasu,N.,LeeSmith-Canter,L.,&Strain,P.S.(2003).Evidence-basedpracticesforyoungchildrenwithautism:Contributionsforsingle-subjectdesignresearch.FocusonAutismandOtherDevelopmentalDisabilities,18(3),166-175https://doi.org/10.1177/10883576030180030401

50. Michie,S.,Johnston,M.,Francis,J.,&Hardeman,W.(2005).Behaviourchangeinterventions:Developingaclassificationsystem.workshoppresentedatthe1stannualconferenceoftheUKsocietyforbehaviouralmedicine.London.

24