RESEARCH Open Access

Improving medication management inmultimorbidity: development of theMultimorbiditY COllaborative MedicationReview And DEcision Making (MY COMRADE)intervention using the Behaviour Change WheelCarol Sinnott1*, Stewart W. Mercer2, Rupert A. Payne3, Martin Duerden4, Colin P. Bradley1 and Molly Byrne5

Abstract

Background: Multimorbidity, the presence of two or more chronic conditions, affects over 60 % of patients inprimary care. Due to its association with polypharmacy, the development of interventions to optimise medicationmanagement in patients with multimorbidity is a priority. The Behaviour Change Wheel is a new approach forapplying behavioural theory to intervention development. Here, we describe how we have used results from areview of previous research, original research of our own and the Behaviour Change Wheel to develop anintervention to improve medication management in multimorbidity by general practitioners (GPs), within theoverarching UK Medical Research Council guidance on complex interventions.

Methods: Following the steps of the Behaviour Change Wheel, we sought behaviours associated with medicationmanagement in multimorbidity by conducting a systematic review and qualitative study with GPs. From themodifiable GP behaviours identified, we selected one and conducted a focused behavioural analysis to explain whyGPs were or were not engaging in this behaviour. We used the behavioural analysis to determine the interventionfunctions, behavioural change techniques and implementation plan most likely to effect behavioural change.

Results: We identified numerous modifiable GP behaviours in the systematic review and qualitative study, from whichactive medication review (rather than passive maintaining the status quo) was chosen as the target behaviour.Behavioural analysis revealed GPs’ capabilities, opportunities and motivations relating to active medication review. Wecombined the three intervention functions deemed most likely to effect behavioural change (enablement, environmentalrestructuring and incentivisation) to form the MultimorbiditY COllaborative Medication Review And DEcision Making (MYCOMRADE) intervention. MY COMRADE primarily involves the technique of social support: two GPs review themedications prescribed to a complex multimorbid patient together. Four other behavioural change techniques areincorporated: restructuring the social environment, prompts/cues, action planning and self-incentives.

Conclusions: This study is the first to use the Behaviour Change Wheel to develop an intervention targetingmultimorbidity and confirms the usability and usefulness of the approach in a complex area of clinical care. Thesystematic development of the MY COMRADE intervention will facilitate a thorough evaluation of its effectiveness inthe next phase of this work.

* Correspondence: csinnott@ucc.ie1Department of General Practice, University College Cork, Cork, IrelandFull list of author information is available at the end of the article

ImplementationScience

© 2015 Sinnott et al. 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.

Sinnott et al. Implementation Science (2015) 10:132 DOI 10.1186/s13012-015-0322-1

BackgroundMultimorbidity, the presence of two or more chronicconditions, affects over 60 % of patients in primary care[1]. In a healthcare system that has evolved around themanagement of single chronic diseases, this presentsmajor challenges to healthcare provision, research andmedical education [2]. In 2014, the US Department ofHealth and Human Services recognised these challengesby stating the need to better equip clinicians in the man-agement of multimorbidity, making specific reference tomedication management [3]. Multimorbidity frequentlyleads to the prescription of multiple long-term medica-tions [4]. The resulting polypharmacy is an independentrisk factor for negative health outcomes such as adverseeffects and drug interactions [5]. For prescribers, thiscreates a tension between keeping the number of medi-cines to a minimum while still prescribing whatevidence-based guidelines advocate as being in the pa-tient’s best interest [6]. This is especially the case forgeneral practitioners (GPs), who must coordinate andoversee the medications prescribed by numerous doctorsinvolved in the care of a multimorbid patient [7].Despite the prevalence of multimorbidity, few inter-

ventions have been developed to improve medicationmanagement in this field to date. A recent systematic re-view, which focussed on interventions to optimise out-comes in patients with multimorbidity in primary care,found only two that specifically addressed medicationmanagement. However, both interventions related to en-hanced involvement of pharmacists, rather than the pre-scribing actions of GPs [8]. Thus, the development ofinterventions to improve GPs’ contribution to medi-cation management in patients with multimorbidityis a priority.In the past, interventions that aimed to change health-

care professionals’ behaviour have resulted in suboptimaleffects, due to a lack of theoretical consideration at thedevelopment stage [9]. The UK Medical Research Coun-cil (MRC) guidance for the development of complex in-terventions in healthcare emphasises the importance ofusing theory in intervention design [10]. However, theMRC document does not put forth any specific sugges-tions on how to do this which leaves intervention de-signers, many of whom are interested in theory only tothe extent that it can help them achieve improvementsin clinical care, with an array of dilemmas [11]. Thelarge pool of available theoretical models means thatcritical theories may be missed, and there is little clarityon how to choose the most appropriate theory for thebehaviour in question [12]. In addition, intervention de-velopers have traditionally had little to guide them onthe specification of intervention content [13].Over the last few years, this gap has been addressed by

an approach known as the Behaviour Change Wheel

(BCW), which explicitly integrates behavioural theorywith the development and description of behaviouralchange interventions [14]. A core feature of the BCW isa theoretical model which is used to conduct an analysisof the behaviour in question. The model is based on thehypothesis that the interaction between one’s capability(C), opportunity (O) and motivation (M) can provide ex-planations for why a particular behaviour (B) is or is notperformed (COM-B). Each of these components can befurther subdivided (Fig. 1). Capability may be physical(the physical skill, strength and stamina) or psycho-logical (the knowledge or psychological skills, strengthor stamina to engage in the necessary mental processes).Opportunity may be physical (afforded by the environ-ment, including resources, locations, time etc.) or social(afforded by interpersonal influences, social cues, andcultural norms that influence the way we think aboutthings). Motivation may be reflective (plans, self-conscious intentions or evaluations) or automatic (reflexresponses, impulses, drive states). The COM-B behav-ioural analysis guides the choice of intervention func-tions (or strategies) most likely to achieve behaviouralchange. Additionally, the intervention functions havebeen linked to a taxonomy of 93 replicable behaviouralchange techniques [15], and those techniques particu-larly suitable for each intervention function have beenhighlighted [14]. Following this structured approachlends transparency to the process of intervention devel-opment and facilitates its subsequent implementationand evaluation [12].Since its original publication in 2011, the BCW has re-

ceived a lot of academic interest, and a number ofgroups have already used it to develop or study the im-plementation of interventions by healthcare profes-sionals [16–19]. To our knowledge, there are nopublished examples of using the BCW to develop a denovo intervention targeted at healthcare professionals inthe complex field of multimorbidity. As the applicationof the BCW may vary according to the setting and targetbehaviour, examples of the generalizability of the ap-proach are required. Furthermore, published examplesof its use will contribute to the ongoing developmentand refinement of the method.In this paper, we describe the development of an inter-

vention to improve medication management in multi-morbidity by GPs, in which we applied the steps of theBCW to enable a more transparent implementation ofthe MRC framework for design and evaluation of com-plex interventions.

MethodsIn the MRC framework, intervention development com-prises three stages: identifying the evidence base, identi-fying and applying appropriate theory to the available

Sinnott et al. Implementation Science (2015) 10:132 Page 2 of 11

(and if necessary, new) evidence and modelling processand outcomes [10]. Like the MRC framework, the BCW[14] also has three broad stages but they involve differ-ent tasks (i.e. understanding the behaviour, identifyingintervention options and identifying content and imple-mentation options) and are subdivided into a furthereight steps (i.e. defining the problem in behaviouralterms, selecting the target behaviour, specifying the tar-get behaviour, identifying what needs to change, identifyappropriate intervention functions, identifying policycategories, identifying behavioural change techniquesand determining the mode of delivery) [14]. As we wereusing the BCW within the overarching framework of theMRC, we mapped the eight BCW steps directly on tothe three development stages of the MRC to en-hance the clarity and generalizability of our approach(see Table 1).

MRC stage 1: identifying the evidence baseTo begin, we reviewed the existing evidence on medica-tion management in multimorbidity and supplementedthis with new evidence in order to clearly define ourproblem of interest and then select and specify the be-havioural target for intervention.

BCW step 1: define the problem in behavioural termsWe searched for relevant published literature, in particu-lar existing systematic reviews, to help us understandthe problems associated with medication management inmultimorbidity in primary care. While we identified tworelevant quantitative reviews [8, 20], we also found anumber of pertinent qualitative studies. Therefore, weconducted a systematic review and synthesis of the rele-vant qualitative evidence, the methods of which havebeen published elsewhere [21].

Fig. 1 The Behaviour Change Wheel

Table 1 Mapping steps of Behaviour Change Wheel to the three stages of intervention development in the UK Medical ResearchCouncil guide on complex interventions in healthcare

MRC development stage [10] BCW steps [14] BCW stages

1. Identify the evidence base 1. Define the problem in behavioural terms 1. Understand the behaviour

2. Select the target behaviour

3. Specify the target behaviour

2. Identify/develop theory 4. Identify what needs to change

5. Identify appropriate intervention functions 2. Identify intervention options

6. Identifying policy categories

3. Model process and outcomes 7. Identifying behavioural change techniques 3. Identify content and implementation options

8. Determine the mode of delivery

Sinnott et al. Implementation Science (2015) 10:132 Page 3 of 11

We addressed the gaps identified from the qualitativesynthesis by conducting a qualitative interview studywith GPs, specifically to generate further information ontheir approaches to prescribing in multimorbidity. Themethods for the interview study have also been pub-lished elsewhere [22].

BCW step 2: select the target behaviourFrom the aggregated qualitative synthesis and interviewdata, we (CS and CB) identified the modifiable GP be-haviours relating to medication management in multi-morbidity and selected one key behaviour to target inour intervention. This judgement was informed by cri-teria set out in the BCW guide which are the likelihoodthat behavioural change would be implemented, thelikely impact of changing the behaviour, the spillover orknock on effect of change on other behaviours and theease with which each behaviour could be measured [14].

BCW step 3: specify the target behaviourOnce the target behaviour was decided, we specified ingreater detail what and who needs to change and whereand when this change should happen.

MRC stage 2: identifying/developing theoryIn the next stage, we used the COM-B (capability, op-portunity, motivation—behaviour) model to develop atheoretical understanding of the target behaviour andguide our choice of intervention functions.

BCW step 4: identify what needs to change to achieve thedesired behaviourWe used the COM-B model to frame our qualitative be-havioural analysis of the qualitative synthesis and inter-view data. We (CS and CB) coded empirical datarelevant to GPs’ psychological and physical capabilities(C), social and physical opportunities (O) and reflectiveand automatic motivations (M) to highlight why GPswere or were not engaging in the target behaviour andwhat needed to change for the target behaviour to beachieved. Where multiple COM-B components were po-tentially relevant to one section of the data, the compo-nent whose definition (as set out in the BCW guide,[14]) best fit the context of our data was chosen. The re-sults of this analysis was presented to the other authorsat a consensus meeting and refined accordingly.

BCW step 5: identify intervention functions to achieve thedesired behaviourThe BCW incorporates a comprehensive panel of nineintervention functions, shown in Fig. 1, which weredrawn from a synthesis of 19 frameworks ofbehavioural-intervention strategies. We determinedwhich intervention functions would be most likely to

effect behavioural change in our intervention by map-ping the individual components of the COM-B behav-ioural analysis onto the published BCW linkage matrices[14]. Each intervention function seen to be potentiallyrelevant to our data was considered in detail. We usedthe affordability, practicability, effectiveness and cost-effectiveness, acceptability, side effects/safety and equity(APEASE) criteria, another component of the BCW ap-proach, to grade the potentially relevant interventionfunctions into first and second line options [14].

BCW step 6: policy categoriesThe BCW also includes matrices which signpost theseven broad policy-level interventions for achieving be-havioural change, shown in Fig. 1. As we were not pri-marily concerned with changing policy in this study, wedid not undertake this step in detail, other than listingthe options that may be relevant to levering our inter-vention in the future.

MRC stage 3: modelling process and outcomesIn this third stage, we specified our intervention contentin more detail and identified an appropriate way ofimplementing the intervention within our context.

BCW step 7: identify behavioural change techniquesThe selected intervention functions represented ourbroad approach to achieving behavioural change, but werequired fine-grained techniques to operationalise thesefunctions. We used the links previously drawn betweenthe BCW and the taxonomy of 93 behavioural changetechniques [14, 23] to list those techniques most fre-quently used with our selected intervention functions.We held an expert panel consensus meeting to reviewthe suitability of each of these techniques, in the light ofour previously collected qualitative data, the context ofthe intervention and by referring to the APEASE criteria.Each member of the panel had expertise in one or moreareas of relevance (clinical pharmacology and prescrib-ing (CB, MD, RP), general practice (CB, CS, MD, RP,SM), behavioural science and intervention design (MB)and multimorbidity (CS, RP, SM)).

BCW step 8: identify mode of deliveryAs we were developing an intervention to be imple-mented by individual GPs, this step (mode of delivery)required explicit consideration of implementation in theheterogeneous setting of general practice. We used theexpert panel consensus to specifically address modellingquestions posed in the MRC framework which werewould it be possible to use this, what subgroup of pa-tients should it be used for, what outcomes should besought and what are the facilitators/obstacles at practicelevel [10]. If multiple implementation options existed,

Sinnott et al. Implementation Science (2015) 10:132 Page 4 of 11

agreement was reached by discussing each option, withreference to the APEASE criteria [14].

ResultsMRC stage 1: identifying the evidence baseBCW step 1: define the problem in behavioural termsWe identified two existing systematic reviews whichwere relevant. Patterson et al. reviewed existing inter-ventions to improve prescribing and polypharmacy inolder patients [20]. Only one of the included studies in-volved GPs and showed that computer decision supportreduced inappropriate drug initiation in primary care[24]. The authors suggested that future polypharmacyinterventions must address the complexity of clinical sit-uations and the individuality of prescribers. Smith et al.reviewed interventions to improve patient outcomes inmultimorbidity in primary care. Two included studiesaddressed medication management but these involvedpharmacists rather than GPs. Here, the authors sug-gested that future interventions should target specificproblems relating to multimorbidity, be integrated intoexisting healthcare systems and be embedded with inter-professional collaboration [8].Our qualitative synthesis included ten studies from

seven countries involving a total of 275 GPs [21]. A keytheme was GPs’ sense of professional isolation in themanagement of multimorbid patients. This emanatedfrom the interplay between four aspects of the manage-ment of patients with multimorbidity: (i) the disorgan-isation and fragmentation of healthcare between primaryand secondary care, (ii) the inadequacy of guidelines andevidence-based medicine for multimorbidity, (iii) chal-lenges in delivering patient-centred, rather thandisease-focused, care and (iv) barriers to shareddecision-making.In the qualitative interview study, we interviewed 20

GPs about 51 multimorbid cases [22]. We found thatGPs responded to clinical dilemmas in multimorbidityby ‘satisficing’, i.e. accepting care that they deemed satis-factory and sufficient for a particular patient, yet ac-knowledging that aspects of that care may not beoptimal. In patients with changing disease trajectories,satisficing was manifested as relaxing targets for diseasecontrol, negotiating compromise with the patient, ormaking ‘best guesses’ about the most appropriate courseof action to take. In multimorbid patients perceived asstable, GPs’ default approach was to ‘maintain the statusquo’ rather than actively rationalise medications.

BCW step 2: select the target behaviourThe numerous modifiable GP behaviours relating tomedication management in multimorbidity are shown inFig. 2. ‘Maintaining the status quo’ was observed in all ofthe qualitative interviews despite best practice guidelines

which state that patients receiving long-term medicinesneed medication reviews at regular intervals. Targetingthis behaviour would likely result in behavioural changeas the qualitative study showed GPs extant discomfortwith it [22]. Furthermore, it would be desirable to seeGPs adopt a less-passive approach to medication man-agement even if it did not always lead to downstreamchanges to medications. There was a high possibility of‘spill over’ of the actions of medication review for multi-morbidity into other prescribing activities. Lastly, chan-ging this behaviour would be relatively easy to measure.We judged that the other modifiable behaviours werenot as attractive: adopting practice protocols would havea big impact and high spillover, but given current finan-cial and staffing pressure on practices would be a diffi-cult organisational change to achieve; relaxing targetsand negotiating compromise may be argued to be appro-priately patient-centred in multimorbidity and trying tochange these, in a healthcare system where progressivelyless appears to be patient-centred, may be resisted byGPs; addressing shared decision-making has merit butrequires interventions targeting GPs’ communicationskills (rather than prescribing) which was not our spe-cific focus.

BCW step 3: specify the target behaviourThe target behaviour was specified as active, purpose-ful medication review instead of passive ‘maintainingthe status quo’ for patients with multimorbidity, to beconducted by GPs, in routine general practice, on aregular basis.

Fig. 2 Modifiable GP behaviours in medication management inmultimorbidity identified in qualitative synthesis [21] and interviewstudy [22]

Sinnott et al. Implementation Science (2015) 10:132 Page 5 of 11

MRC stage 2: identifying/developing theoryBCW step 4: identify what needs to change to achieve thedesired behaviourWe used COM-B to identify GPs’ capabilities (C), oppor-tunities (O) and motivations (M) for engaging, or notengaging, in active medication review. The themes thatemerged from this analysis are shown in Table 2, with il-lustrative quotes from the qualitative synthesis and theinterview study. For example, GPs adopted a passive ap-proach to medication management due to their uncer-tainty (lack of psychological capability) about whichmedications were most valuable in patients with multi-morbidity, especially given the absence of satisfactoryguidelines in this field. Insufficient time within the con-sultation led to a lack of physical opportunity to reviewmedications. GPs also found medication review difficultbecause of a cultural milieu which holds that treatmentfor chronic disease is lifelong (lack of social opportun-ity). This was especially the case if the patient had beencompliant with their medications for many years. ManyGPs had developed a habitual response to ‘not rock theboat’ in patients with multimorbidity, an approach whichinvolved not making changes to medications unlessthere was a pressing reason to do so. This response wasreinforced by their experiences of the negative conse-quences of stopping or changing medications for pa-tients with multimorbidity in the past (automaticmotivation). GPs’ reflective motivations against medica-tion review included the opportunity cost of using theirprofessional time for this purpose and a fear of negativeconsequences from rationalising medications. GPs alsohad motivations to review medications which includedimproving patient outcomes, reassuring themselves thatthey are delivering best care, and guarding againstmedico-legal repercussions.

BCW step 5: identify intervention functionsWe found that all nine intervention functions listed inthe BCW were relevant to our behavioural analysis.Additional file 1 shows our assessment and grading ofeach intervention function using the APEASE criteriainto first and second line options. The three interventionfunctions most relevant for our intervention were en-ablement, environmental re-structuring and incentivisa-tion. The relationship between the components of theCOM-B behavioural analysis and the three selectedintervention functions are shown in Table 2.

BCW step 6: policy categoriesThe broad policy options, signposted by the BCW matri-ces as being potentially useful for achieving behaviouralchange, were communication/marketing, service provisionpolicy, legislation, guidelines and regulation.

MRC stage 3: modelling process and outcomesBCW step 7: identify behavioural change techniquesFrom the taxonomy of 93 behavioural change tech-niques, we listed the techniques most frequentlyused to deliver the three intervention functions wehad selected [14, 23]. The resulting 32 potentiallyrelevant techniques are listed in Additional file 2. Inthe expert panel, we reviewed how each techniquecould be applied to the context of medication man-agement in multimorbidity. The panel’s choice oftechniques was influenced principally by the keyfindings of the qualitative studies: GPs’ sense of iso-lation in the management of multimorbid patientsrevealed in the qualitative synthesis [21], and GPs’lack of certainty and efforts to ‘share the onus of re-sponsibility’ seen in the interview study [22]. Thus,we focused on options that would enhance GPs’means of professional support. Although enhancedcommunication between GPs and pharmacists is be-ing investigated in other healthcare systems [25, 26],it is not currently an option in Irish general practicedue to the lack of community pharmacists. Similarly,communication between GPs and the specialists in-volved in multimorbidity patient care was seen inboth qualitative studies to be fraught with poor ac-cess and a single-disease approach. A useful sourceof support for some GPs in the interview study wastheir GP colleagues. These interactions occurred onan informal basis within practices and were notablefor their ready accessibility and generalist nature[22]. We were unaware of any work exploring col-laborative decision-making between GPs in multi-morbidity, so focused on this approach. From thelist of 32, we considered which techniques wouldpragmatically facilitate collaborative decision-makingbetween GPs. Additional file 2 shows how manywere quickly eliminated as not being relevant to thecontext or purpose of the intervention. The fivetechniques eventually selected as ‘active ingredients’were social support (practical), restructuring the so-cial environment, use of prompts/cues, action plan-ning and self-incentives. The definition of eachtechnique and qualitative data to support their selec-tion are shown in Table 2. The combination and inte-gration of each technique into the overall intervention,named MultimorbiditY COllaborative Medication Re-view And DEcision Making (MY COMRADE), isshown in Table 3.

BCW step 8: identify mode of deliveryIn the expert panel meeting, we then formulated anintervention implementation plan. Four specific as-pects of implementation were reviewed, and the vari-ous options considered for each aspect are fully

Sinnott et al. Implementation Science (2015) 10:132 Page 6 of 11

Table 2 Behavioural analysis, selected intervention functions and behavioural change techniques, referencing empirical data fromthe qualitative synthesis (QS) and the interview study (IS)

Sinnott et al. Implementation Science (2015) 10:132 Page 7 of 11

described in Additional file 3. In summary, the fol-lowing implementation plan was formulated:

1. What prompts should be used to guide medicationreview in MY COMRADE?After reviewing eight different prescribing tools andchecklists (listed in Additional file 3), it was agreedthat a modified version of the seven prompts in theNOTEARS [27] checklist for medication reviewwould be used to cue the review.

2. How should GPs choose which patients to reviewusing MY COMRADE?After reviewing multiple options (see Additional file3), it was agreed that GPs should choose patientsprescribed ten or more regular medicines or four ormore medicines with at least one other complicatingfactor (i.e. meets criteria for potentiallyinappropriate prescribing, at risk of a well-recognised drug-drug interaction, has poor adher-ence or receiving end-of-life or palliative care), inline with recommendations from the Kings Fund re-port on Polypharmacy and Medication Optimisation[28].

3. How should the behavioural change technique“action planning” be operationalised?One of the behavioural change techniques, actionplanning, specifically relates to implementation andwas selected to account for the wide variety ofstructures and systems that occur in generalpractice. Each GP will be given clear guidance onhow to tailor MY COMRADE to suit their practice.This will involve asking them to choose a particularday, time of the day and office in which to do thereview. They will decide on the number of cases toreview in one sitting, and the GP pairs that willconduct reviews within a practice. In advance of

trialling MY COMRADE, GPs will be asked toconsider what they envision as problematic for itsimplementation, and how these problems could betackled, knowing their own practice.

4. How should the intervention be evaluated?The initial evaluation will focus on interventionimplementation (i.e. did medication review takeplace?). The behavioural change techniques andother causal or contextual mechanisms associatedwith behavioural change will be determined usingqualitative methods. If MY COMRADE is shown tobe effective, future evaluations will assess otherhealth outcomes such as the number of/type ofmedication changes made and changes in rates ofhealthcare utilisation.

DiscussionThis paper describes the systematic, structured develop-ment of an intervention to improve medication manage-ment for multimorbid patients by GPs. The interventionis called MY COMRADE. It is, to our knowledge, thefirst intervention directed at the management of multi-morbidity in primary care, developed by using the Be-haviour Change Wheel to clearly implement theframework of the MRC guide on complex interventions.MY COMRADE involves collaborative decision-

making by two GPs who support each other in thereview of medications prescribed to a complex multi-morbid patient, guided by cues which relate to safeprescribing. The broad functions of the intervention(enablement, environmental restructuring and incenti-visation) are theoretically based. These functions willbe achieved using five specific behavioural changetechniques: social support (practical), restructuringthe social environment, use of prompts/cues, actionplanning and self-incentives. The technique of colle-gial social support is a crucial feature of our interven-tion, which we expect will greatly enable GPs’capabilities in conducting active medication reviews.It may be particularly important in de-prescribingmedications or prioritising patient-centred rather thandisease-focused care in multimorbidity which arechallenging aspects of medication management, notleast because of the fear of litigation which this inter-vention may now help ameliorate.

Comparison with other workSince its publication in 2011, the BCW has been used inthe development of interventions targeting healthcare pro-fessionals in a variety of ways. For example, Alexander etal. used COM-B to understand barriers and enablers topreventative health examinations for young children inAustralian general practice, with a view to designing an

Table 3 Description of final intervention

The final intervention is called MultimorbiditY COllaborative MedicationReview And DEcision Making (MY COMRADE)

It involves the following (relevant behavioural change techniques are inbrackets):

GPs will be asked to schedule protected time for themselves and one oftheir GP colleagues to conduct the collaborative medication review andenter this time into the practice appointment book. They will be askedto choose a day/time/office that suits them best and decide how manypatient cases to review in one sitting (action planning). The GPs willchoose multimorbid patients from their caseload and in the scheduledreview time will review medications, supported by their GP colleague(social support and restructuring social environment). The medicationreview will be cued by the seven prompts described in the NO TEARS[27] medication checklist (prompts and cues). GPs will be asked torecord recommendations for medication change that arise from thereview in the patient’s notes to allow them to discuss these with thepatient during their next consultation. After completing the review, GPswill award themselves continuing professional development points: onepoint for each cumulative hour of the activity completed (self-incentives).

Sinnott et al. Implementation Science (2015) 10:132 Page 8 of 11

implementation intervention to increase the conduct ofthese examinations [16]. They did not describe later stepsof the BCW, such as choice of intervention functions, anddid not describe in detail how their implementation inter-vention would look. In contrast, we used the BCW tohighlight areas for improvement in professional practiceand then develop an intervention targeted to these areas,rather than simply increasing the implementation of apre-existing intervention.Murphy et al. used COM-B to develop a capacity-

building programme to enhance pharmacists’ roles inmental health care [19]. This group felt that implemen-tation processes must be prioritised during the earlystages of intervention development, and they wove theor-ies of behavioural change and implementation together inan iterative way. While we agree that implementationshould be considered at all stages of development work,we did not find it necessary to use a specific implementa-tion framework. The initial steps of the BCW revealedmultiple areas for improvement in GPs’ professional prac-tice. Once one had been chosen, the remaining steps ofthe BCW involved developing an implementation inter-vention to enhance the performance of this desired behav-iour. Additionally, by incorporating the behaviour changetechnique of action planning, implementation was expli-citly integrated into our intervention. Action planning re-quires an individual GP to plan the frequency, durationand intensity of the planned intervention activity [14].Thus, rather than a prescriptive implementation strategy,action planning will allow each GP to adapt the interven-tion for use within their own practice. The variation in im-plementation, as well as fidelity to other behaviouralchange techniques, will be evaluated in the next phase ofthis work and will help to inform the debate on optimalapproaches to implementation planning in interventiondevelopment.

Strengths and weaknessesWe began this work with the broad aim of developingan intervention to improve medication managementin multimorbidty, but we did not have a predefinedidea of what the intervention would be at the outset.Adhering to the guidance of the MRC by using a the-oretical approach, which was chosen a priori, lent dir-ection, structure and transparency to this process inmultiple ways.First, the MRC states the need to identify the evi-

dence base and supplement this with new evidence ifnecessary. In doing this, we generated much neededdata on the management of medications in multimor-bidity, increased our understanding of the problematicareas experienced by GPs and revealed how they cur-rently respond to these difficulties. Second, we then

used this empirical data to directly influence the de-velopment of the intervention. Following the steps ofthe BCW allowed us to develop a list of options forbehavioural change and to clarify what we were, andwhat we were not, trying to achieve. Third, we bene-fitted from using the links between the BCW modeland the taxonomy of behavioural change techniques.The taxonomy highlighted novel strategies for behav-ioural change, many of which we would heretoforenot have considered. Although only five techniquesare ultimately included in the description of the finalintervention, many of the others influenced other as-pects of intervention development and the implemen-tation strategy.Despite the highly systematic and structured approach

of the BCW, there are challenges associated with its useand it is not a magic bullet for intervention develop-ment. For example, the researcher must make a series ofsubjective and pragmatic decisions throughout theprocess. These ‘real life’ decisions can seem at odds withthe scientific approach. To counter this and to improvethe transparency and generalizability of our methods, werecorded in detail the multiple options available to us ateach step of the BCW and expanded on why optionswere or were not taken.Furthermore, the multiple steps of intervention devel-

opment involved a lengthy process: from the beginningof our systematic review to the final refinements of theintervention spanned almost 3 years. Such a prolongedcourse must be factored in by those pursuing and fund-ing evidence-based intervention development. Otherintervention developers have used a ‘top-down’ approachof applying classical behavioural theories such as socialcognitive [29] or control theory [30] to inform theirchoice of intervention functions and behavioural changetechniques. In contrast, we employed a ‘bottom-up’ ap-proach to theory development in which the frameworkof the BCW guided our use of existing evidence andour own qualitative explorations. This led to anintervention which was logical and practical yet stilltheoretically based.In addition to the COM-B, the BCW also includes an

optional, more detailed framework for behavioural analysisknown as the Theoretical Domains Framework [14]. Aftercompleting our intervention development, we conducteda validation using the Theoretical Domains Framework(see Additional file 4) which reassuringly demonstratedsimilar associations between our qualitative data, and ourchosen intervention functions and behavioural changetechniques.

Implications for future researchWe used the BCW as a lens for viewing GP behaviour,understanding what needed to shift and determining

Sinnott et al. Implementation Science (2015) 10:132 Page 9 of 11

how this shift could be achieved. Our experience con-firms the usefulness and generalizability of this ap-proach. Multimorbidity presents many challenges toGPs, particularly relating to the conflicts betweenpatient-centred and disease-focused care but the BCWapproach was not hampered by these complexities.Based on our experience, the method is potentially ap-plicable to intervention developers across disciplines aslong as sufficient contextual and empirical data exists orcan be generated.Throughout this study, we adhered to the ‘less is more’

maxim of intervention design [14]. We could have takena more complex multi-faceted approach, such as incorp-orating other stakeholders, i.e. pharmacists or specialists.Instead, we adopted the recommendations from the sys-tematic review by Smith et al. that changes targetingspecific problems are more likely to be effective [8].Smaller changes can be achieved, sustained and builtupon in future interventions, and substantial behaviouralchange is more likely to result from the aggregation ofthese smaller changes [14]. We applied the same tenetsto our assessment of outcomes—rather than initiallylooking at downstream effects such as changes in pre-scribing, we will concentrate first on proximal changessuch as implementation of the intervention. Once weare assured that it is acceptable, feasible and leads to be-havioural change, we can assess outcomes in prescribingsafety and polypharmacy at a later stage.To date, there is limited evidence available on which

behavioural change techniques are most effective in spe-cific settings. We expect that characterising the activecomponents in the MY COMRADE intervention usingthe taxonomy of behavioural change techniques [15] willaid implementation and replication of the intervention.The clear specification of the intervention will also facili-tate a thorough evaluation of the impact of the selectedbehavioural change techniques and will help to informevidence-based strategies for intervention developmentin the future.In this study, we did not undertake the sixth step

of the BCW relating to policy options in detail. How-ever, if the intervention is shown to be effective inour ongoing feasibility and pilot work, scaling-up ofthe intervention will require greater consideration ofthe external context of healthcare policy and wide-spread implementation.

ConclusionThis paper describes the development of an interventionto improve medication management in multimorbidity byGPs. The intervention, which is called MY COMRADE, isbased on purposively collected data on behaviour incontext and a novel approach to intervention design, theBehaviour Change Wheel. While the Behaviour Change

Wheel is not a magic bullet for intervention design, thispaper confirms the usability and usefulness of this ap-proach in a complex area of clinical care. The systematic,transparent approach used in the development of the MYCOMRADE intervention will facilitate a thorough evalu-ation of its effectiveness in the next phase of this work.

Additional files

Additional file 1: BCW step 5: identify intervention functions usingAPEASE criteria. (DOCX 26 kb)

Additional file 2: BCW step 7: identify behavioural changetechniques. (DOCX 24 kb)

Additional file 3: BCW step 8: identify mode of delivery usingexpert panel. (DOCX 33 kb)

Additional file 4: Validation of the chosen intervention functionsand behavioural change techniques using the theoretical domainsframework. (DOCX 32 kb)

AbbreviationsBCW: Behaviour Change Wheel; COM-B: capability, opportunity,motivation—behaviour. Refers to a model comprised of threecomponents—capability, opportunity, motivation—that are necessary for agiven behaviour to occur; GP: general practitioner; MRC: Medical ResearchCouncil.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsCS designed the study, conducted data analysis and wrote initial draft of thepaper. CB contributed to study design, data analysis and revising themanuscript. MB contributed to study design, analysis and revising themanuscript. SW, MD and RP contributed to data analysis, judgements onintervention design and revising the manuscript. All authors have given finalapproval of the version to be published and agree to be accountable for allaspects of the work.

AcknowledgementsThis work was supported by the Irish Research Council New Foundationsscheme and the Health Research Board (HRB/NSAFP/2011/3 to CS). Thefunding bodies had no role in the study design, collection, analysis orinterpretation of data. The support of the South East GP Training Programmein facilitating Dr. Sinnott's clinician academic fellowship is gratefullyacknowledged.

Author details1Department of General Practice, University College Cork, Cork, Ireland.2General Practice and Primary Care, Institute of Health and Wellbeing,University of Glasgow, Glasgow, UK. 3Cambridge Centre for Health ServicesResearch, Institute of Public Health, University of Cambridge, Cambridge, UK.4Centre for Health Economics and Medicines Evaluation, Bangor University,Bangor, UK. 5Health Behaviour Change Research Group, School ofPsychology, National University of Ireland, Galway, Ireland.

Received: 12 June 2015 Accepted: 9 September 2015

References1. Glynn LG, Valderas JM, Healy P, Burke E, Newell J, Gillespie P, et al. The

prevalence of multimorbidity in primary care and its effect on healthcare utilization and cost. Fam Pract. 2011;28:516–23.

2. Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B.Epidemiology of multimorbidity and implications for health care,research, and medical education: a cross-sectional study. Lancet.2012;380:37–43.

Sinnott et al. Implementation Science (2015) 10:132 Page 10 of 11

3. Bayliss EA, Bonds DE, Boyd CM, Davis MM, Finke B, Fox MH, et al.Understanding the context of health for persons with multiple chronicconditions: moving from what is the matter to what matters. Ann FamMed. 2014;12:260–9.

4. Payne RA, Avery AJ, Duerden M, Saunders CL, Simpson CR, Abel GA.Prevalence of polypharmacy in a Scottish primary care population. Eur JClin Pharmacol. 2014;70:575–81.

5. Tinetti ME, Bogardus Jr ST, Agostini JV. Potential pitfalls of disease-specific guidelines for patients with multiple conditions. N Engl J Med.2004;351:2870–4.

6. Boyd CM, Darer J, Boult C, Fried LP, Boult L, Wu AW. Clinical practiceguidelines and quality of care for older patients with multiple comorbiddiseases. JAMA. 2005;294:716–24.

7. Smith SM, O’Kelly S, O’Dowd T. GPs’ and pharmacists’ experiences ofmanaging multimorbidity: a ‘Pandora’s box’. Br J Gen Pract. 2010;60:285–94.

8. Smith SM, Soubhi H, Fortin M, Hudon C, O’Dowd T. Interventions forimproving outcomes in patients with multimorbidity in primary care andcommunity settings. Cochrane Library. 2012;4:CD006560.

9. Michie S, Johnston M, Abraham C, Lawton R, Parker D, Walker A. Makingpsychological theory useful for implementing evidence based practice: aconsensus approach. Qual Saf Health Care. 2005;14:26–33.

10. Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, Petticrew M.Developing and evaluating complex interventions: the new MedicalResearch Council guidance. BMJ. 2008;337:a1655.

11. Davidoff F, Dixon-Woods M, Leviton L, Michie S. Demystifying theory and itsuse in improvement. BMJ Qual Saf. 2015;24(3):228–38.

12. Michie S, van Stralen MM, West R. The behaviour change wheel: a newmethod for characterising and designing behaviour changeinterventions. Implement Sci. 2011;6:42.

13. Michie S, Abraham C, Eccles MP, Francis JJ, Hardeman W, Johnston M.Strengthening evaluation and implementation by specifying components ofbehaviour change interventions: a study protocol. Implement Sci. 2011;6:10.

14. Michie S, Atkins L, West R. The behaviour change wheel: a guide todesiging interventions. Great Britain: Silverback Publishing; 2014.

15. Michie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W, et al.The behavior change technique taxonomy (v1) of 93 hierarchically clusteredtechniques: building an international consensus for the reporting of behaviorchange interventions. Ann Behav Med. 2013;46:81–95.

16. Alexander KE, Brijnath B, Mazza D. Barriers and enablers to delivery ofthe Healthy Kids Check: an analysis informed by the TheoreticalDomains Framework and COM-B model. Implement Sci. 2014;9:60.

17. Fleming A, Bradley C, Cullinan S, Byrne S. Antibiotic prescribing in long-term care facilities: a qualitative, multidisciplinary investigation. BMJOpen. 2014;4:e006442.

18. English M. Designing a theory-informed, contextually appropriateintervention strategy to improve delivery of paediatric services inKenyan hospitals. Implement Sci. 2013;8:39.

19. Murphy AL, Gardner DM, Kutcher SP, Martin-Misener R. A theory-informed approach to mental health care capacity building forpharmacists. Int J Ment Health Syst. 2014;8:46.

20. Patterson SM, Hughes C, Kerse N, Cardwell CR, Bradley MC.Interventions to improve the appropriate use of polypharmacy for olderpeople. Cochrane Database Syst Rev. 2012;5:CD008165.

21. Sinnott C, Mc Hugh S, Browne J, Bradley C. GPs’ perspectives on themanagement of patients with multimorbidity: systematic review andsynthesis of qualitative research. BMJ Open. 2013;3:e003610.

22. Sinnott C, Hugh SM, Boyce MB, Bradley CP. What to give the patientwho has everything? A qualitative study of prescribing formultimorbidity in primary care. Br J Gen Pract. 2015;65:e184–191.

23. Colquhoun H, Leeman J, Michie S, Lokker C, Bragge P, Hempel S, et al.Towards a common terminology: a simplified framework ofinterventions to promote and integrate evidence into health practices,systems, and policies. Implement Sci. 2014;9:51.

24. Tamblyn R, Huang A, Perreault R, Jacques A, Roy D, Hanley J, et al. Themedical office of the 21st century (MOXXI): effectiveness ofcomputerized decision-making support in reducing inappropriateprescribing in primary care. CMAJ. 2003;169:549–56.

25. Improving the management of patients with multimorbidity in generalpractice. In Current Controlled Trials web site. http://www.controlled-trials.com/ISRCTN06180958; 2014.

26. Avery AJ, Rodgers S, Cantrill JA, Armstrong S, Cresswell K, Eden M, et al. Apharmacist-led information technology intervention for medication errors(PINCER): a multicentre, cluster randomised, controlled trial and cost-effectiveness analysis. Lancet. 2012;379:1310–9.

27. Lewis T. Using the NO TEARS tool for medication review. BMJ. 2004;329:434.28. Duerden M, Avery A, Payne R. Polypharmacy and medicines optimisation.

Making it safe and sound. The King’s Fund: London, UK; 2013.29. Lakshman R, Griffin S, Hardeman W, Schiff A, Kinmonth AL, Ong KK. Using

the Medical Research Council framework for the development andevaluation of complex interventions in a theory-based infant feedingintervention to prevent childhood obesity: the baby milk intervention andtrial. J Obes. 2014;2014:646504.

30. Gray NM, Allan JL, Murchie P, Browne S, Hall S, Hubbard G, et al. Developinga community-based intervention to improve quality of life in people withcolorectal cancer: a complex intervention development study. BMJ Open.2013:3(4).

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit

Sinnott et al. Implementation Science (2015) 10:132 Page 11 of 11