UBC Faculty Research and Publications

Mechanisms of action of an implementation intervention in stroke rehabilitation: a qualitative interview… Connell, Louise A; McMahon, Naoimh E; Tyson, Sarah F; Watkins, Caroline L; Eng, Janice J Sep 30, 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12913_2016_Article_1793.pdf [ 424.28kB ]
JSON: 52383-1.0318067.json
JSON-LD: 52383-1.0318067-ld.json
RDF/XML (Pretty): 52383-1.0318067-rdf.xml
RDF/JSON: 52383-1.0318067-rdf.json
Turtle: 52383-1.0318067-turtle.txt
N-Triples: 52383-1.0318067-rdf-ntriples.txt
Original Record: 52383-1.0318067-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessMechanisms of action of animplementation intervention in strokerehabilitation: a qualitative interview studyLouise A. Connell1*, Naoimh E. McMahon1, Sarah F. Tyson2, Caroline L. Watkins1 and Janice J. Eng3AbstractBackground: Despite best evidence demonstrating the effectiveness of increased intensity of exercise after stroke,current levels of therapy continue to be below those required to optimise motor recovery. We developed andtested an implementation intervention that aims to increase arm exercise in stroke rehabilitation. The aim of thisstudy was to illustrate the use of a behaviour change framework, the Behaviour Change Wheel, to identify themechanisms of action that explain how the intervention produced change.Methods: We implemented the intervention at three stroke rehabilitation units in the United Kingdom. A purposivesample of therapy team members were recruited to participate in semi-structured interviews to explore theirperceptions of how the intervention produced change at their work place. Audio recordings were transcribedand imported into NVivo 10 for content analysis. Two coders separately analysed the transcripts and coded emergentmechanisms. Mechanisms were categorised using the Theoretical Domains Framework (TDF) (an extension of theCapability, Opportunity, Motivation and Behaviour model (COM-B) at the hub of the Behaviour Change Wheel).Results: We identified five main mechanisms of action: ‘social/professional role and identity’, ‘intentions’,‘reinforcement’, ‘behavioural regulation’ and ‘beliefs about consequences’. At the outset, participants viewedthe research team as an external influence for whom they endeavoured to complete the study activities. Thestudy design, with a focus on implementation in real world settings, influenced participants’ intentions toimplement the intervention components. Monthly meetings between the research and therapy teams werecentral to the intervention and acted as prompt or reminder to sustain implementation. The phased approachto introducing and implementing intervention components influenced participants’ beliefs about the feasibilityof implementation.Conclusions: The Behaviour Change Wheel, and in particular the Theoretical Domains Framework, were usedto investigate mechanisms of action of an implementation intervention. This approach allowed for consideration of arange of possible mechanisms, and allowed us to categorise these mechanisms using an established behaviour changeframework. Identification of the mechanisms of action, following testing of the intervention in a number of settings,has resulted in a refined and more robust intervention programme theory for future testing.Keywords: Stroke, Implementation, Behaviour change, Mechanisms of action, Qualitative design* Correspondence: laconnell@uclan.ac.uk1College of Health & Wellbeing, University of Central Lancashire, Preston PR12HE, UKFull list of author information is available at the end of the article© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Connell et al. BMC Health Services Research  (2016) 16:534 DOI 10.1186/s12913-016-1793-8BackgroundThe time it takes for research to translate into clinicalpractice is unacceptably long [1]. Systematic reviews andclinical guidelines have been devised in efforts to bridgethis evidence-practice gap. However, such publicationsalone are not enough to initiate and sustain a change inthe day-to-day practices of clinicians [2], and the under-pinning evidence is often criticised for not being reflect-ive of how interventions are delivered in the real clinicalcontext. There remains a lack of understanding aboutcontextual factors influencing stoke rehabilitation prac-tice. For example, despite best evidence demonstratingthe effectiveness of increased intensity of exercise afterstroke, current levels of therapy continue to be belowthose required to optimise motor recovery [3, 4]. Severaltherapy interventions have been developed to addressthis problem with respect to recovery of the arm afterstroke but important aspects of these interventions areoften modified when implemented in routine care [5].Thus the extent to which they achieve their aim of in-creasing intensity of exercise in real world rehabilitationsettings is unclear. We have developed an implementa-tion intervention, underpinned by implementation the-ory, that aims to increase arm exercise in strokerehabilitation by changing the behaviour of therapists.The intervention is underpinned by formative researchon an evidence-based arm rehabilitation intervention,the Graded Repetitive Arm Supplementary Programme(GRASP) [6]. GRASP is a self-directed hand and arm ex-ercise programme which is taught and monitored by atherapist, but carried out by the patient with the supportof their family/carer where possible. GRASP is notmeant to replace existing therapy services, rather to aug-ment current therapy, adding opportunities for morepractice. Similar to existing therapy interventions,GRASP involves a complex implementation chain influ-enced by interactions between patients, therapists andthe wider rehabilitation environment. The fidelity to theintervention in clinical settings had been shown to bevariable [5].Development of the implementation intervention wasguided by the Behaviour Change Wheel (BCW) [7], the de-tails of which have been reported elsewhere [8]. In brief, weworked collaboratively with a therapy team at a local strokeunit and other stakeholders, establishing an active partner-ship in an intervention development group. Structured dis-cussions were undertaken to (i) understand the problem(i.e. intensity of arm exercise in the stroke rehabilitationunit) and to identify target behaviours that would be amen-able to change, (ii) design intervention components thatcould change behaviours through established behaviourchange techniques and (iii) pilot and refine the developedintervention components. The resulting intervention wascalled PRACTISE (Promoting Recovery of the Arm: Clin-ical Tools for Intensive Stroke Exercise). It consists of face-to-face meetings between the research team and therapyteams, and materials to aid implementation using estab-lished behaviour change techniques, a novel aspect of thisimplementation study compared with many previous re-habilitation studies. PRACTISE aims to address four targetbehaviours of therapists: (i) screening patients for suitabilityfor supplementary self-directed arm exercise, (ii) provisionof arm exercises, (iii) involving family/carers in assistingwith exercises and (iv) monitoring and reviewing exercises.Our original programme theory for PRACTISE wasdeveloped using the Capability, Opportunity, Motivationand Behaviour model (COM-B) at the hub of the BCWand is summarised in Table 1. The behaviour changetechniques underpinning the intervention componentswere identified using the Behaviour Change TechniqueTaxonomy v1 (numbers shown correspond with how the93 techniques are clustered into 16 groups in thetaxonomy) [9].We intended to influence physical and social opportun-ity to perform the target behaviours by making it easier fortherapy teams to provide supplementary self-directed armexercises to patients, and by making the provision of armexercises a priority in the stroke rehabilitation unit.Monthly meetings between the research team and therapyteams provided an opportunity to discuss emergentbarriers to implementation and to identify context appro-priate solutions, which were intended to maximisecommitment to the study and implementation of theintervention. Materials to support implementation in-cluded a screening tool and exercise pack to make itquicker and easier to provide exercises to suitable patients.Table 1 Original proposed PRACTISE intervention components and mechanisms of action using the behaviour change wheel;capability, motivation, behaviour model; and BCT taxonomy (v1)Intervention components Determinant of behaviour from COM-B Behaviour change techniques from BCTTv1a [13]Screening toolPRACTISE exercise packPhysical Opportunity Due to time constraints more efficient ways ofperforming the target behaviours were needed4.1 Instruction on how to perform the behaviour3.2 Social support (practical)Team meetingsAudit toolSocial Opportunity Getting upper limb rehabilitation higher up onthe agenda was needed through managerialsupport and team engagement1.2 Problem solving1.4 Action planning1.9 Commitment2.3 Self-monitoring of behaviourBCTTv1: Behaviour Change Technique Taxonomy v1aBehaviour change techniques are numbered based on the Behaviour Change Technique TaxonomyConnell et al. BMC Health Services Research  (2016) 16:534 Page 2 of 10The audit tool provided a method of self-monitoring forperformance of the target behaviours at a service level.We anticipated that this would work in two ways. Firstly,as a weekly reminder about what needed to be done. Sec-ondly, if there was a discrepancy between what the teamsproposed to do and what they actually did, this would behighlighted and act as an incentive to improve for the fol-lowing week.We implemented PRACTISE in two additional strokeunits to explore the feasibility and acceptability of the inter-vention to staff and patients, along with examining the waysin which the intervention facilitated change. To develop ef-fective implementation interventions researchers are beingurged not just to establish that an intervention works, butalso to identify and explain the specific ways in which itworks i.e. the mechanisms of action [10]. Mechanisms havebeen defined as “hidden but real” and as elements “of rea-soning and reactions of agents in regard to the resourcesavailable in a given context” [11]. The term mechanism hasbeen conceptualised and operationalised differently in theliterature, the most well-known approach perhaps beingthe context-mechanism-outcome configurations used inrealist evaluation [12]. For the purposes of our study weconsidered mechanisms to be those less observable “things”that happen in the black box between behaviour changetechniques and components, and observed outcomes.Although work is ongoing to establish a method forlinking behaviour change techniques to mechanisms ofaction [13] there is not yet an established approach foridentifying and reporting the mechanisms of action ofbehaviour change or the components of implementationinterventions. The aim of this study was to use the BCWto identify mechanisms of action and provide a rich ex-planation as to how our implementation interventionsupported change at a site level. We consider the emer-gent mechanisms in light of our original programmetheory to present a refined intervention programme the-ory for future testing.MethodsImplementation of the interventionThe characteristics of the participating sites are shownin Table 2. A detailed report on the implementation ofthe intervention across the three sites (Sites A, B & C),and staff and patients’ perceptions of the feasibility andacceptability of the intervention will be reported in detailseparately. The extent to which the target behaviourschanged in the three stroke rehabilitation units over thecourse of the study is shown in Table 3.PRACTISE resulted in change in at least two of the fourtarget behaviours at each site. In site A, therapy assistantsprovided assistance and supervision to patients while per-forming their upper limb exercises, rather than family/carers. At Site B, resignation of senior staff and ongoingservice re-organisation limited the extent to which wecould progress through the phased implementation ap-proach in the six month study period. Site C, acted as thedevelopment site for the intervention and therefore had alonger “embedding period”. Changes in all target behav-iours were achieved at this site and were sustained as a re-sult of including upper limb therapy practice in aninternal departmental audit. The least change was seen inmonitoring and progressing of exercises across the threestroke units due to a short length of stay.Study designQualitative interview study.Theoretical frameworkWe used a directed content analysis approach withbehaviour change theory used as guidance for initialcodes [14].Participant selectionAll physiotherapists, occupational therapists, therapymanagers and therapy assistants in the participating siteswere involved in the embedding of PRACTISE. ATable 2 Characteristics of participating sitesSite information Site A Site B Site COrganisation General hospital General hospital General hospitalNumber of stroke beds 23 24 24Patients admitted from Emergency department Hyper-acute stroke ward Hyper-acute stroke wardAverage length of stay 18.5 days Missing 23 daysWeekday therapy input Target of 45 mins therapy perdiscipline per dayTarget of 45 mins of each therapy per day Target of 45 mins of each therapy per dayWeekend therapy input Reduced Saturday service(prioritise chest physiotherapyand new patients)No service on SundaysReduced Saturday service (prioritise chestphysiotherapy and new patients)No service on SundaysNone routinelyStaffing (WTE, when full) PT: 6.0OT: 6.0Assistants: 3.0PT: 3.8OT: 4.0Assistants: 4.5PT: 3.1OT: 2.8Assistants: 1.7Connell et al. BMC Health Services Research  (2016) 16:534 Page 3 of 10purposive sample of these participants were invited totake part in interviews to explore their perceptions ofhow the intervention produced change, or not, at theirsite during the study. The rationale for conducting inter-views over the six month study period was to learnabout the processes of change, and how these may havedeveloped from the beginning to the end of the study.Between two and three participants were interviewedduring each visit based on their availability. Participantswere only interviewed once over the course of the study.On average, interviews took approximately 45 min. Weceased recruiting therapy team members once data sat-uration was reached (i.e. no new themes were emerging)or all members of the therapy team had participated.SettingWe used a six month phased approach to implementa-tion in three stroke rehabilitation units in the NorthWest of England (the stroke unit where the interventionwas developed and two additional stroke units). Imple-mentation was guided by the target behaviours (i.e. start-ing with screening of patients before progressing toprovision of arm exercises) and commenced at Sites Aand B in October 2014. Site C acted as the developmentsite for the intervention from December 2013 to June2014.Data collectionSemi-structured face-to-face interviews were conductedto explore therapists’ perceptions of how the interven-tion produced, or failed to produce, change were con-ducted by LC and NM on site in quiet spaces and atconvenient times for the interviewees. Where possible,interviews were conducted in private offices, but some-times they were conducted in quiet corners of publicspaces, e.g. the hospital canteen due to space limitations.Normalisation Process Theory (NPT) was used to de-velop an interview guide (Additional file 1). NPT is asociological toolkit to understand the work that is doneto implement and embed complex interventions inhealthcare settings [15]. Particular emphasis was placedon probing questions that encouraged participants to re-flect on what supported change throughout the stages ofimplementation. Interviews were audio-recorded and allparticipants provided written informed consent prior tothe interview. Field notes were made after each site visitto document the following: observations, the content ofmonthly meetings; ad hoc discussions with therapists;additional contacts (e.g. email) between meetings andreasons for these; and informal discussions on the pro-gress of the study by therapists and managers. Thesedata were summarised at the end of data collectionperiod to provide more detailed insight into the processof implementation and possible mechanisms, providinga method of triangulation.Data analysisAudio recordings were transcribed and imported intoNVivo 10 for content analysis. Interview transcripts werecoded by LC and NM using predetermined codes basedon the Theoretical Domains Framework (TDF) [16]. TheTDF is an extension of the Capability, Opportunity andMotivation model (COM-B) at the hub of the BCW.Codes were compared between researchers and non-fitting items discussed. An agreement was reached onwhere the mechanisms fit within the TDF, with any fur-ther points of contention discussed with all the authorsand agreement sought. Emergent mechanisms were dis-cussed with study participants to ensure that the datahad been accurately interpreted and to provide oppor-tunity for clarification of preliminary findings. The finalcoding process involved free coding of text where partic-ipants provided rich and insightful reflections as to howand why the intervention produced change.Synthesis of Results to produce mechanisms of action forPRACTISEFollowing the final coding process, the research teammet to synthesise the results by listing the interventioncomponents and to relate these to the findings from theperceived mechanisms of action. Discrepancies betweenthe determinants of behaviour as assigned a priori in thedevelopment stage using the COM-B model, and pos-sible mechanisms of action as identified by the TDFwere discussed and agreement made about how theintervention is understood to work. Issues with this willbe considered in the discussion.Research teamLC and NM are both chartered physiotherapists withexperience of qualitative research methods. Both holdfull-time research positions at a UK university workingon a National Institute for Health Research (NIHR)funded project to develop a clinically feasible and struc-tured upper limb exercise programme for use inNational Health Service (NHS) stroke rehabilitationunits. They built a relationship with the therapy teamsTable 3 Change in target behavioursTarget behaviour Site A Site B Site CScreening patients for suitability forsupplementary self-directed arm exercise✓ ✓ ✓Provision of arm exercises ✓ ✓ ✓Involving family/carers in assisting with exercises O ✘ ✓Monitoring and reviewing exercises ✘ ✘ ✓✓ = most change in performanceO = implemented alternative change than proposed in the intervention✘ = least change in performanceConnell et al. BMC Health Services Research  (2016) 16:534 Page 4 of 10at each of the sites throughout the study but were notknown to the participants beforehand. The therapyteams were informed at the outset that the purpose ofthe research was to explore the feasibility of imple-menting practice change in stroke rehabilitation set-tings as opposed to demonstrating the effectiveness ofa specific intervention. ST, CW and JE are stroke re-habilitation researchers. JE developed GRASP, the inter-vention from which this study emerged, and conductedthe randomised trial confirming its effectiveness [6]. Adirected approach to content analysis was undertakenas it makes explicit that the research team had experi-ence of implementation theories and were not workingfrom a naïve perspective, and fitted with the aim of thestudy to use existing theory (the TDF) to try and un-pick mechanisms of action of the intervention.ResultsCharacteristics of interview participantsTwenty-three therapy team members were interviewed:8 physiotherapists, 11 occupational therapists and fourtherapy assistants (Table 4). Six participants had beenqualified for five years or less and 10 participants hadmore than five years’ experience specifically workingwith people with stroke. Of the qualified staff, two werejunior (NHS band 5), 14 were senior (NHS band 6 or 7),and 3 were team leads/therapy manager. Only one staffmember was male. The breakdown of demographic byparticipant, or identifiers within the included quotes, hasnot been included to protect the anonymity of theparticipants.Mechanisms of actionUsing the TDF, we identified five mechanisms thatcould explain how, or why, PRACTISE produced theobserved changes. These included: ‘social/professionalrole and identity’; ‘intentions’; ‘reinforcement’; ‘behav-ioural regulation’, and ‘beliefs about consequences’.Definitions of these TDF mechanisms are providedbelow. For some emergent mechanisms, clear linkswith specific components of the intervention, or be-haviour change techniques could be identified butquite often, participants discussed how their involve-ment in the study as a whole resulted in change attheir site.Social/professional role and identityThis domain relates to a coherent set of behaviours anddisplayed personal qualities of an individual in a socialor work setting. Participants accredited their team’s en-gagement with the study to two factors relating to pro-fessional roles. Firstly, they viewed the research team asan external influence for whom they wanted to ensurethe required work was completed. This links to the con-structs of professional credibility and identity within theTDF. Secondly, they valued their relationship with theuniversity, which gave an impetus to ensure they deliv-ered the required work. The social identity, and howtherapists related to the research team influenced theirbehavior.Site A, PT04: “Even if you had someone sort ofdriving it forward within the team, I don’t knowwhether it has quite the same effect as an externalforce that gives you that sort of…“yeah I really shoulddo that before they come in”…But that’s no bad thingactually until it’s at the point where it’s embedded.”Site C, OT11: “For me personally, I think there was abig impetus to do it because it was linked with UCLaninitially. You kind of have something that you’reaiming towards.”IntentionsIntentions relate to a conscious decision to perform abehaviour or a resolve to act in a certain way. Partici-pants accredited their intent to engage with the study toits design. At the outset, we stressed that the purpose ofthe study was to test the feasibility of implementingPRACTISE at their work setting and that all feedback orsuggested revisions would be welcome. This meant ther-apists did not feel threatened and were willing to movefrom contemplation and preparation to action. Thistheme emerged particularly strongly at Site B where thetherapy team were implementing PRACTISE during aservice re-organisation, and hence had difficulties per-forming the target behaviours. When therapy teams werereassured that capturing all of these experiences andchallenges was worthwhile for the research, they feltunder less pressure to perform all target behaviours con-sistently, and as a consequence persevered with thestudy processes. Emphasising that PRACTISE could, andshould, fit with ‘real life working’ seemed to resonatewith participants and was very much in contrast to theirpast research experiences.Site B, OT09: “I think because there’s been so manyproblems in the team, there was talk about peoplewanting to withdraw but then the meeting that I wentto where LC said “we know it’s not been very good,and we know it’s all been tricky for you, don’t worryTable 4 Interview participants across sitesSite Total PT OT AssistantA 12 5 6 1B 6 2 3 1C 5 1 2 2Connell et al. BMC Health Services Research  (2016) 16:534 Page 5 of 10about it and that’s useful information to us”. I thinkthat really helped and everyone was like “oh, that’salright then” because it felt like we were failing beforeand it was like a stress that we couldn’t manage andthat we weren’t doing what was asked of us but thenwith LC saying that everyone was like “oh brill”…”Site A, OT06: “I just think that it’s been a really nicerelaxed project to be involved in. We’ve never feltpressurised into getting the results and suchlike…Sometimes when you’re involved in research it’s…you’ve just got to get the numbers in and it becomes areal sort of turn off in some respects.”At the development site, Site C, upper limb therapyinput was used for the team’s internal annual audit,which acted as a driving force to sustain implementationeven after the research team’s involvement had come toan end.Site C, PT08: “You guys obviously took a step backso that was less of a drive really to keep it going, butthen because the project was linked with ourdepartmental audit, that then gave us anotherdeadline that we had to work towards…So myself andone of the OT’s had to get our act together again, togather that data for a slightly different reason but thathas made the rest of the team, and now the new staffthat have come in, more aware of that process and ithas become embedded again within our practice Iwould say.”ReinforcementReinforcement relates to increasing the probability of aresponse by arranging a dependent relationship betweenthe response and a given stimulus. For the PRACTISEintervention, the active involvement of the researchersand the regular team meetings provided reinforcementto perform the target behaviours, and meant that therewas recognition amongst peers if behaviours were per-formed, and conversely negative consequences of report-ing that behaviours weren’t being undertaken.Site A, PT04: “As I said, it’s been good that you guyshave been coming because I think it’s kept us thinkingabout it and it’s also moved it forward…”Site C, OT11: “I do think it’s been quite valuable tohave consistent input from the people introducing thetreatment activities; not just to amend it or whateverbut to keep the momentum going.”The challenge of maintaining momentum when imple-menting new treatment approaches was highlighted inthe interviews. Participants discussed past experiences ofcolleagues who have returned to work after attendingContinuing Professional Development (CPD) events.New ideas from these events were often very well re-ceived when first introduced but with time tended to fallby the wayside.Site A, PT02: “From experience, things are very hotwhen they’re new and then kind of tail off. And everynow and again, someone will remember it and try andpick it up but I think this (regular meetings) seems tobe quite a nice refresher.”Behavioural regulationBehavioural regulation is anything aimed at changingobjectively measured actions. The PRACTISE interven-tion relates to the constructs of self-monitoring and ac-tion planning. The purpose of the audit tool wasspecifically to facilitate self-monitoring performance oftarget behaviours. Participants confirmed that the audittool in weekly meetings acted as a reminder to keep upthe PRACTISE activities. However, they viewed the toolmore as research data than as a method of monitoringoverall service performance. Site C was an exception, asthey were using the data collected to conduct an internalaudit in their department. Therapists also discussed howthe team meetings acted as a prompt to plan who wouldbe responsible for each of the target behaviours for eachpatient.Site A, PT03: “Yeah and I kind of feel like, if I was towork anywhere else I’d find something similar usefulso…each week we go “OK, who needs a PRACTISEprogramme?” and having our tick boxes, becauseotherwise I think it’s very easy to forget about thesetools.”Site A, PT04: “I can see where it does sort of help.You know it’s nice to look back and it’s nice to go“OK you have done that”…but I don’t think we’vemade any attempt to really look at it as a team. I don’tthink that’s what’s driven us forward which is why Idon’t know whether it would make a difference if wecontinued it or not, or whether we would see it as justanother bit of paperwork that needs looking at anddoing.”Beliefs about consequencesThis domain relates to acceptance of the truth, reality,or validity about outcomes of a behaviour in a givensituation. In this instance, beliefs about consequences re-fers to the consequences of implementing PRACTISEfor the therapy teams, rather than the consequences forpatients. At the outset, therapists were understandablyconcerned about the feasibility of implementing some-thing new with already constrained resources. However,as the study progressed, therapists’ attitudes towards theConnell et al. BMC Health Services Research  (2016) 16:534 Page 6 of 10value of the intervention seemed to change whereby itwas no longer seen as an added burden but an integralpart of their therapy that brought reward.Site A, PT05: “I think it’s definitely been worthwhile,I think it’s really changed what we’re doing on theunit… It’s something additional that we don’tnecessarily have time to focus on in therapy sessions,so it gives an extra opportunity for more therapythroughout the day…I think despite all those kind ofinitial thoughts it’s…now we’re at the point where it’sjust part of what we do.”Site A, PT03: “At first I think it was difficult becauseit was like an extra thing to do so you had your owntreatment plan of what you wanted to do with apatient then you’re like oh, I have to do thisPRACTISE as well so to begin with I kind of saw it asa separate thing…I don’t now, I must say it’s given meloads of good little ideas with for exercises with them.I’d say that I definitely incorporate it more into thetherapy session and as opposed to an add-on.”PRACTISE intervention components and mechanisms ofactionA refined list of the PRACTISE intervention compo-nents and mechanisms of action which form the re-fined programme theory are presented in Table 5.More detailed description on the components of theintervention are provided with a related expansion inthe behaviour change techniques delivered throughthese components. The TDF mechanisms of action arepresented along with their relationship to the COM-Bmodel categories. The content has been organised tobe readable and show some connections between ac-tions, techniques and mechanisms. However, in realitythese overlap and mechanisms can act as precursorsor successors to each other. Although ‘environmentalcontext and resources’ did not emerge as a particularlystrong mechanism in the interviews we have includedit here as a potential explanation as to how techniquessuch as ‘adding objects to the environment’ may pro-duce change. It is perhaps the case, that physical re-structuring of an environment is not a mechanism initself, but that it works by bringing about a change inother mechanisms: for example intentions. This willbe considered further in the discussion. Conversely,the mechanism ‘beliefs about consequences’ was notattributable to individual intervention components oractivities but instead emerged as a reflection on thestudy process as a whole. It has therefore not been in-cluded in Table 5.DiscussionWe have demonstrated use of the Behaviour Change Wheelto identify five possible mechanisms of action of an imple-mentation intervention. These included ‘social/professionalTable 5 Refined PRACTISE intervention components and and BCT taxonomy (v1); mechanisms of action using the behaviourchange wheel; capability, motivation, behaviour model; and TDFMonthly meetings between research teamand therapy teamsIntervention components Behaviour change techniques fromBCTTv1 [13]Determinant of behaviourfrom COM-BMechanisms of action from TDFMonthly meetings between research andtherapy teams7.1 Prompts/cues Automatic motivation ReinforcementBeliefs about consequencesResearch team from local university 10.5 Social incentive Reflective motivation Social/professional role and identifyPhased approach focussing on feasibilityof implementation4.1 Instruction on how to performthe behaviour8.7 Graded tasksReflective motivation IntentionsIdentified barriers to performing targetbehaviours and developed strategies toovercome them1.2 Problem solving1.4 Action planning3.2 Social support (practical)12.5. Adding objects to the environmentReflective motivationPhysical opportunityIntentionsBehavioural regulationIntervention components Behaviour change techniques fromBCTTv1 [13]Source of behaviour fromCOM-BMechanisms of action from TDFScreening tool & exercise packMaterials provided to assist performanceof the target behaviours12.5 Adding objects to the environment Physical opportunity Environmental context andresourcesAudit toolTherapy teams asked to documentperformance of the target behavioursand provide feedback by research team2.2 Feedback on behaviour2.3 Self-monitoring of behaviour7.1 Prompts/cuesPsychological capability Behavioural regulationBCTTv1 behaviour change technique taxonomy v1, BCW behaviour change wheel, TDF theoretical domains frameworkConnell et al. BMC Health Services Research  (2016) 16:534 Page 7 of 10role and identity’, ‘intentions’, ‘reinforcement’, ‘behaviouralregulation’ and ‘beliefs about consequences’. In the originalintervention development, we hypothesised that implemen-tation of the intervention would occur through changes inphysical and social opportunity. However, the emergentmechanisms most often related to reflective and automaticmotivation (TDF domains of ‘social/professional role andidentity’, ‘intentions’, ‘reinforcement’, and ‘beliefs about conse-quences). At the outset, the therapy teams’ motivation toengage in the study activities was attributed to the monthlyvisits with the research team to discuss progress, which mo-tivated them to complete the study activities as much aspossible in the interim periods. The phased approach to im-plementation and the focus on feasibility sustained motiv-ation throughout. Furthermore, a collaborative workingrelationship with the research team that encouraged teamsto provide feedback on the intervention, and how it couldbe refined or revised, gave therapists’ autonomy to adaptintervention components to fit with their local context.This contrasted with their prior experiences of implement-ing strict research trial protocols. Once the therapy teamsperceived the intervention to be part of their routine work,and of some value, motivation was driven less by a feelingof having to implement the intervention and more by want-ing to implement it. This fits with other stroke rehabilita-tion implementation research, which highlights the needfor active management strategies and close collaborationwith stakeholders [2].The importance of exploring and reporting the mecha-nisms of action of interventions has been highlighted inmethodological and reporting guidance including theTemplate for Intervention Description and Replication(TIDieR) [17] and the Medical Research Council guid-ance for process evaluations [10]. However, as yet thereare few examples of how to operationalise this in imple-mentation research. To date, the TDF has been used to(i) explore barriers and facilitators to performing targetbehaviours e.g. [18, 19], (ii) guide intervention develop-ment [20–22] and (iii) describe intervention contentboth prior to implementation and through retrospectiveanalysis [23–25]. There are limited examples where thisframework has been used to unpick the mechanisms ofaction of components of interventions. Where examplesdo exist, identification of the mechanisms of action isoften oriented around hypothesized or expected mecha-nisms of action by mapping behaviour change tech-niques to domains of the TDF, rather than to qualitativeanalysis of mechanisms of action as experienced by therecipients of interventions [23, 26] thus limiting compar-isons that can be made with our study.Methodological considerationsIn this study we set out to identify mechanisms of actionthat supported change. This is not to present an overlypositive picture of the intervention, or to ignore explana-tions as to why some changes did not occur. Barriers tochange often related to the feasibility of performing tar-get behaviours within contextual constraints. Therefore,by developing a thorough understanding of the mecha-nisms that promoted change, it may be possible to iden-tify prerequisite contextual conditions that allow thesemechanisms to be activated or thrive.The TDF enabled categorising of the mechanisms ofaction, which had both strengths and limitations. It gavemore detail than the COM-B, as it is an extension of thismodel, though arguably the definitions are more obtuse.Although definitions for the domains of the TDF areprovided, it was not always easy to determine in whichcategories the emergent mechanisms would be bestplaced. Some of the terminology used in the TDF wasfound to be confusing and jargon heavy, which may notbe helpful to some users of the research e.g. clinicians.This is perhaps an irony within implementation re-search, that the frameworks in themselves are not alwaysuser friendly. In addition, there is an underlying assump-tion that the mechanisms are static, whereas in realitythey may be different for different people and across dif-ferent contexts. The extent to which identified mecha-nisms were linked was also unclear. For example, itwould seem plausible that the mechanisms under ‘so-cial/professional role and identify’ could be consideredantecedents to changes in ‘intentions’.As mentioned in the results section, it was difficult todetermine whether discussions around ‘environmentalcontext and resources’ could be conceptualised as mech-anisms in themselves, or whether adding objects to theenvironment (e.g. screening tool, exercise pack) triggeredmechanisms such as ‘behavioural regulation’. Whenreviewing the domains of the TDF, it is clear that theyinclude a mix of personal characteristics (e.g. social/pro-fessional role and identify), cognitive processes (e.g. in-tentions), responses or reactions to components andtechniques (e.g. reinforcement) and physical changes(e.g. environmental context and resources). Further workmay be needed to establish greater consistency abouthow the term ‘mechanism’ is conceptualised and opera-tionalised using these frameworks.Strengths and limitationsA strength of this study is the use of an establishedframework to categorise and describe the mechanisms ofaction, which facilitated consistent definitions and ter-minology. In turn, this allows for improved understand-ing of the content of the intervention and the ways inwhich it may produce change. This will also allow for ro-bust comparisons to be made across studies for furthertesting and development of behaviour change theory andframeworks. Conversely, this use of a directed contentConnell et al. BMC Health Services Research  (2016) 16:534 Page 8 of 10approach has some inherent limitations in that data isanalysed with potential bias, and a desire to fit all re-sponses to pre-determined categories even if the fit isnot perfect. We attempted to minimise this through dis-cussions with the research team and participants, tocheck interpretation of the results. However, a limitationof this study may be the reliability of the analysisprocess. Neither coder has undertaken formal training inuse of the behaviour change framework or its taxonomy.Despite our best efforts to use terms and definitions ac-curately and consistently, some categories were broadand overlapping. Consequently, some mechanisms couldbe categorised in alternative ways. Furthermore, like allinterview studies, we were reliant on participants’ will-ingness and ability to fully reflect on, and articulate,their experiences. The depth of reflection, and richnessof explanation reported in the transcripts, was variedand meant that heavy reliance was put on the more de-tailed and insightful interview accounts to extract themechanisms of action, rather than the transcripts de-scribing more practical aspects of implementation. Forthis reason, it is important to clarify that the mecha-nisms presented here are intended to be useful explana-tions as to how the components of the intervention mayproduce change. Alternative mechanisms may have beenpresent that were not captured in the interview tran-scripts, and it is likely that participants will experiencedifferent mechanisms and in different ways. This is alimitation of behaviour change taxonomies, which aregrounded in psychological theory, with contextual as-pects given little consideration. This contrasts with, forexample, realist methodologies. It was clear in this study,that interventions work through different mechanisms indifferent environments, and it is not yet clear how to de-scribe or account for this using the TDF.LC and NM facilitated implementation at each site andalso conducted the interviews. Participants may have beeninclined to provide favourable responses to the interviewers’questions (i.e. a social desirability bias [27]) but it wasstressed throughout that the purpose of the study was tolearn about the process of implementing the interventionto encourage participants to be candid in relaying their ex-periences. Participants were interviewed at different timepoints through the implementation process, and repeatedinterviews were not possible due to limiting time demandson the staff involved. However, it has been shown that re-call of the processes involved in implementation can belimited [5] therefore interviewing people at different pointsalong the implementation process was important.ConclusionsThere is increasing emphasis being placed on establishingnot only if an intervention works, but how it works. Wehave illustrated the use of the Behaviour Change Wheel,and in particular the use of the Theoretical DomainsFramework, to investigate mechanisms of action of an im-plementation intervention. This approach allowed for con-sideration of a range of possible mechanisms, and allowedus to categorise these mechanisms using an establishedbehaviour change framework. Identification of the mecha-nisms of action, following testing of the intervention in anumber of settings, has resulted in a refined and more ro-bust intervention programme theory for future testing.Additional fileAdditional file 1: Appendix 1. Interview guide. Details of the interviewguide used, developed using NPT. (DOCX 18 kb)AbbreviationsBCTTv1: Behaviour change technique taxonomy v1; BCW: Behaviour changewheel; COM-B: Capability opportunity motivation and behaviour;CPD: Continuing professional development; GRASP: Graded repetitive armsupplementary programme; NPT: Normalisation process theory;NIHR: National Institute for Health Research; NRES: National research ethicsservice; TDF: Theoretical domains framework; TiDIER: Template forintervention description and replication; PRACTISE: Promoting recovery ofthe arm: clinical tools for intensive exerciseAcknowledgementsWe would like to acknowledge the contribution of the three therapy teamsthat were involved in this implementation study for their time, hard workand willingness to share their experiences of being part of the study.Availability of data and materialsThe transcripts will not be published to preserve participants’ privacy but areavailable from authors upon request.Authors’ contributionsLC, CW and JE made substantial contributions to conception and design.Data collection and data analysis was carried out by LC and NM. All authorswere involved in drafting the manuscript and revising it critically forimportant intellectual content. All authors have given final approval of theversion to be published and take public responsibility for appropriateportions of the content. All authors agree to be accountable for all aspectsof the work in ensuring that questions related to the accuracy or integrity ofany part of the work are appropriately investigated and resolved.Competing interestsThe authors declare that they have no competing interests.LC and NM are funded by a National Institute for Health Research CareerDevelopment Fellowship. This article presents independent research fundedby the National Institute for Health Research (NIHR). The views expressed arethose of the author(s) and not necessarily those of the NHS, the NIHR or theDepartment of Health.Consent for publicationNot applicable.Ethical approval and consent to participateThis study was approved by the National Research Ethics Service (NRES), RECnumber 14/NW/1087, IRAS project ID: 157255. All participants providedwritten informed consent prior to participating in the study.Author details1College of Health & Wellbeing, University of Central Lancashire, Preston PR12HE, UK. 2Stroke & Vascular Research Centre, School of Nursing, Midwifery &Social Work, Jean McFarlane Building, University of Manchester, Oxford Rd,Manchester M13 9PL, UK. 3Department of Physical Therapy, University ofBritish Columbia, 212-2177 Wesbrook Mall, Vancouver, BC V6T 1Z3, Canada.Connell et al. BMC Health Services Research  (2016) 16:534 Page 9 of 10Received: 2 February 2016 Accepted: 24 September 2016References1. Davis D, Evans M, Jadad A, Perrier L, Rath D, Ryan D, et al. The case forknowledge translation: shortening the journey from evidence to effect. BMJ.2003;327:33–5.2. Walker MF, Fisher RJ, Korner-Bitensky N, McCluskey A, Carey LM. From whatwe know to what we do: translating stroke rehabilitation research intopractice. Int J Stroke. 2013;8:11–7.3. Lang C, MacDonald J, Gnip C. Counting repetitions: an observational studyof outpatient therapy for people with hemiparesis post-stroke. J NeurolPhys Ther. 2007;31:3–10.4. Lang C, MacDonald J, Reisman D, Boyd L, Kimberley T, Schindler-Ivens S, etal. Observation of Amounts of Movement Practice Provided During StrokeRehabilitation. Arch Phys Med Rehabil. 2009;90:1692–8.5. Connell L, McMahon N, Harris J, Watkins C, Eng J. A formative evaluation ofthe implementation of an upper limb stroke rehabilitation intervention inclinical practice: a qualitative interview study. Implement Sci. 2014;9:90.6. Harris JE, Eng JJ, Miller WC, Dawson AS. A Self-Administered GradedRepetitive Arm Supplementary Program (GRASP) Improves Arm FunctionDuring Inpatient Stroke Rehabilitation A Multi-Site Randomized ControlledTrial. Stroke. 2009;40:2123–8.7. Michie S, van Stralen MM, West R. The behaviour change wheel: a newmethod for characterising and designing behaviour change interventions.Implement Sci. 2011;6:42.8. Connell LA, McMahon NE, Redfern J, Watkins CL, Eng JJ. Development of abehaviour change intervention to increase upper limb exercise in strokerehabilitation. Implement Sci. 2015;10:34.9. Michie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W, etal. The behavior change technique taxonomy (v1) of 93 hierarchicallyclustered techniques: Building an international consensus for the reportingof behavior change interventions. Ann Behav Med. 2013;46:81–95.10. Moore GF, Audrey S, Barker M, Bond L, Bonell C, Hardeman W, et al. Processevaluation of complex interventions: Medical Research Council guidance.BMJ. 2015;350:h1258.11. Lacouture A, Breton E, Guichard A, Ridde V. The concept of mechanismfrom a realist approach: a scoping review to facilitate its operationalizationin public health program evaluation. Implement Sci. 2015;10:153.12. Pawson R, Tilley N. Realistic evaluation. London: Sage; 1997.13. UCL Centre for Behaviour Change. Research. http://www.ucl.ac.uk/behaviour-change/applied-research. Accessed 30 November 2015.14. Hsieh HF, Shannon SE. Three approaches to qualitative content analysis.Qual Health Res. 2005;15:1277–88.15. May C, Finch T. Implementing, Embedding, and Integrating Practices: AnOutline of Normalization Process Theory. Sociol. 2009;43:535–54.16. Cane J, O’Connor D, Michie S. Validation of the theoretical domainsframework for use in behaviour change and implementation research.Implement Sci. 2012;7:37.17. Hoffmann TC, Glasziou PP, Boutron I, Milne R, Perera R, Moher D, et al.Better reporting of interventions: template for intervention description andreplication (TIDieR) checklist and guide. BMJ. 2014;348:g1687.18. Cadogan CA, Ryan C, Francis JJ, Gormley GJ, Passmore P, Kerse N, et al.Improving appropriate polypharmacy for older people in primary care:selecting components of an evidence-based intervention to targetprescribing and dispensing. Implement Sci. 2015;10:161.19. Bain E, Bubner T, Ashwood P, Van Ryswyk E, Simmonds L, Reid S, et al.Barriers and enablers to implementing antenatal magnesium sulphate forfetal neuroprotection guidelines: a study using the theoretical domainsframework. BMC Pregnancy Childbirth. 2015;15:176.20. French SD, Green SE, O’Connor DA, McKenzie JE, Francis JJ, Michie S, et al.Developing theory-informed behaviour change interventions to implementevidence into practice: a systematic approach using the TheoreticalDomains Framework. Implement Sci. 2012;7:38.21. Tavender EJ, Bosch M, Gruen RL, Green SE, Michie S, Brennan SE, et al.Developing a targeted, theory-informed implementation intervention usingtwo theoretical frameworks to address health professional andorganisational factors: A case study to improve the management of mildtraumatic brain injury in the emergency department. Implement Sci.2015;10:74.22. Porcheret M, Main C, Croft P, McKinley R, Hassell A, Dziedzic K.Development of a behaviour change intervention: a case study on thepractical application of theory. Implement Sci. 2014;9:42.23. Steinmo S, Fuller C, Stone SP, Michie S. Characterising an implementationintervention in terms of behaviour change techniques and theory: The‘Sepsis Six’ clinical care bundle. Implement Sci. 2015;10:111.24. McCullough AR, Ryan C, O’Neill B, Bradley JM, Elborn JS, Hughes CM.Defining the content and delivery of an intervention to Change AdhereNceto treatment in BonchiEctasis (CAN-BE): a qualitative approach incorporatingthe Theoretical Domains Framework, behavioural change techniques andstakeholder expert panels. BMC Health Serv Res. 2015;15:342.25. Little EA, Presseau J, Eccles MP. Understanding effects in reviews ofimplementation interventions using the Theoretical Domains Framework.Implement Sci. 2015;10:90.26. Kolehmainen N, Francis JJ. Specifying content and mechanisms of changein interventions to change professionals’ practice: an illustration from theGood Goals study in occupational therapy. Implement Sci. 2012;7:100.27. Adams AS, Soumerai SB, Lomas J, Ross-Degnan D. Evidence of self-reportbias in assessing adherence to guidelines. Int J Qual Health Care.1999;11:187–92.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Connell et al. BMC Health Services Research  (2016) 16:534 Page 10 of 10


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items