Optimizing a digital intervention for managing blood pressure in stroke patients using a diverse sample: Integrating the person‐based approach and patient and public involvement

Abstract Background Having a stroke or transient ischaemic attack increases the risk of a subsequent one, especially with high blood pressure (BP). Home‐based BP management can be effective at maintaining optimal BP. Objective To describe the optimization of a digital intervention for stroke patients and the value of participant diversity, using the person‐based approach (PBA) and integral patient and public involvement (PPI). Setting and participants Stroke patients recruited from primary care and community settings, and health‐care professionals in primary care, in England and Ireland. Design Three linked qualitative studies conducted iteratively to develop an intervention using the PBA, with integral PPI. Intervention The BP: Together intervention, adapted from existing BP self‐monitoring interventions, is delivered via mobile phone or web interface to support self‐monitoring of BP at home. It alerts patients and their clinicians when a change in antihypertensive medication is needed. Findings Feedback from a diverse range of participants identified potential barriers, which were addressed to improve the intervention accessibility, feasibility and persuasiveness. Easy‐to‐read materials were developed to improve usability for patients with aphasia and lower literacy. The importance of including family members who support patient care was also highlighted. Feedback messages regarding medication change were refined to ensure usefulness for patients and clinicians. Discussion Input from PPI alongside qualitative research with a diverse study sample allowed the creation of a simple and equitable BP management intervention for stroke patients. Patient involvement Two PPI co‐investigators contributed to design, conduct of study, data interpretation and manuscript preparation; community PPI sessions informed early planning. Study participants were stroke patients and family members.


| INTRODUC TI ON
Stroke is the 4th leading cause of death in the UK, with more than 100 000 strokes or transient ischaemic attacks (TIA) occurring every year. 1 Almost two-thirds of people who have had a stroke leave hospital with a disability and/or suffer from limb impairment.
Approximately one third are affected by aphasia, a complex language and communication disorder resulting from damage to the language centres of the brain. 1 People from Black and minority ethnic groups (BAME), as well as those living in socially deprived areas, are at increased risk of hypertension-related stroke. [2][3][4] Having a stroke increases the risk of having a subsequent one, especially for those with high blood pressure (BP). Evidence suggests that effective BP control may be even more important for the prevention of secondary stroke than primary stroke 5 and current guidelines recommend 6 that antihypertensive treatment be 'increased as quickly as tolerated' for people who have had a stroke or TIA (hereafter referred to as 'stroke patients'). However, BP control in stroke patients is known to be suboptimal 7 and is estimated to increase the risk of recurrent stroke by 1/3 for every 10 mmHg increase in systolic BP. 8 BP control is closely linked to BP management, which can often be suboptimal due to inertia from both patients and clinicians. [9][10][11] In the UK and Western Europe, only around 14%-26% of patients with uncontrolled hypertension have their medication increased. 12 Any intervention hoping to improve BP management in stroke patients therefore needs to understand and address both patient-and clinician-level barriers.

| Developing interventions
The Medical Research Council (MRC) guidance on developing and evaluating complex interventions helped to standardize some key elements involved in intervention development 13 such as identifying the evidence base and developing a strong and appropriate theoretical underpinning for an intervention. Paying attention to context is key: what works in one setting for one subgroup of the population may be less effective, or even harmful, elsewhere for other subgroups. Contextual factors require sensitive examination, not only when evaluating and implementing an intervention into the 'real world' but also from its inception. 14 Newer guidance, specifically for intervention development, 15 recognizes that context significantly impacts the appropriateness and success of any intervention, emphasizes the importance of involving stakeholders throughout development and recommends refining the intervention in response to stakeholder feedback.
Working with the target population during intervention development enables a detailed understanding of users' psychosocial contexts, and the identification of potential barriers to the target behaviours. 15 The person-based approach (PBA) 16 provides a clear process for achieving this. This iterative approach moves cyclically between data collection involving in-depth qualitative research with the target user population, analysis to identify behavioural barriers, modifications to the intervention and then further data collection. 17 Ideally, participants should be sampled carefully to reflect the diversity of people affected by the particular health condition. In reality, however, some groups are consistently under-represented in research, [18][19][20] which can lead to their perspectives being overlooked and their needs not addressed. 21,22 Another complementary approach to prioritizing users' needs is to work closely with patient and public involvement (PPI) contributors. This enables members of the public to be actively involved in research, 23 ideally at every stage of the research cycle. 24 Although PPI does not provide 'data' like the qualitative interviews using PBA, PPI contributors provide valuable oversight on the research questions, design, methods, conduct and even interpretation of findings, ensuring these are in line with patient priorities and needs. Muller et al 25 have highlighted the value of both PPI and PBA during all stages of intervention development. PPI enables a deep, longer-term involvement of stakeholders in the conception, development and implementation of the intervention. Using PBA allows individuals who are not personally invested in the intervention to objectively inform its development from their varied perspectives and contexts.

| Aim of this paper
This paper aims to describe the optimization of a digital intervention for stroke patients, particularly focusing on the value of involving a diverse range of participants, using the person-based approach with integral patient and public involvement. Patient involvement: Two PPI co-investigators contributed to design, conduct of study, data interpretation and manuscript preparation; community PPI sessions informed early planning. Study participants were stroke patients and family members.

K E Y W O R D S
blood pressure, community-based recruitment, digital health, hypertension, patient and public involvement, person-based approach, qualitative research, seldom-heard groups, selfmonitoring, stroke

| ME THODS
This work was informed by the MRC guidance 13 and a more up-todate framework of actions for intervention development. 15 In the early stages of planning, the study team reviewed published research from other BP self-monitoring trials [26][27][28] and drew on existing theories 29 to inform the protocol and the logic model (Appendix S1). This paper describes the qualitative research and PPI that informed the development of the BP:Together intervention.

| Design
Researchers conducted three qualitative studies (Figure 1), using PBA methods with patients and health-care professionals alongside on-going PPI (see Section 2.3): 1. Study 1: Think-aloud, face-to-face interviews 30 with stroke patients (alone or with a family member) to explore perceptions of prototype intervention materials, identify possible barriers to engagement and modify as appropriate; 2. Study 2: Retrospective interviews with stroke patients (telephone or face-to-face, according to patient preference) who had used the prototype intervention for a week, to explore their experiences and identify further barriers to engagement; 3. Study 3: Focus groups with clinicians and support staff to explore feasibility and concerns about implementing the intervention in primary care.

Studies 1 and 3 were conducted in London and the Thames
Valley area in England, and in the Mid-west and Midlands in Ireland.
Study 2 was only conducted in England as the technical systems supporting the prototype intervention were not in place for Ireland.

| The BP:Together intervention
The BP:Together intervention was adapted from the TASMINH4 28 and HOME BP 17 interventions but with design features and content specifically to facilitate its use following stroke/TIA. It supports self-monitoring of blood pressure (BP) and enables patients to share BP readings taken at home with their clinician for the initiation of planned medication changes when average readings are above target. Figure 2 shows the system overview and interactions between the intervention modules.
When BP:Together is trialled in primary care, clinicians will complete a mandatory online training module about the intervention before recruiting any patients. Patients randomized to the intervention will receive in-person training from a nurse and an information booklet (the 'BP:Together booklet') for on-going support at home.
The training session includes registering the patient on the digital intervention, the patient practising taking their own BP readings, and sending them via the website, SMS message or app, according to patient preference. The rationale for providing the intervention across three different interfaces is explained below: 1. The website enables access to the intervention via a desktop or laptop computer, which has a larger display and easier navigation via a mouse and keyboard, compared with using a smartphone or tablet.
2. For people who use smartphones or tablets, the app option provides an immersive experience allowing quick and easy access to information, notifications and BP records (graphs, filtered message history and help pages).
3. The SMS option does not require the use of a smartphone and is likely to be appreciated for its simplicity. It is ideal for participants who might be put off accessing the intervention if they had to use a smartphone or tablet. The SMS approach offers the essential basic functionality needed to interact with the intervention: the patient sends their readings in a toll-free SMS message and receives SMS reminders and feedback on their readings each time.
Once the patient has been trained, the clinician reviews the patient's current antihypertensive medication and makes a three-step plan for dose increases or additional drugs, to be actioned during the 12-month programme.
The patient then receives reminder prompts in the format of their choice (email, app notification or SMS message) to measure their BP, and receives instant acknowledgements when readings are sent. The patient receives overall feedback based on their average, using an F I G U R E 1 BP:Timeline for the three studies during intervention development algorithm based on the NICE guidelines for stroke/TIA patients. 6 If their average BP is above target (systolic BP above 125 mmHg), an email prompts the clinician to initiate the next medication change in their three-step plan. 31 In anticipation of clinical inertia arising from clinicians' concerns that patients might not accept increases to their medication, 10,32 the BP:Together intervention also invites patients to send a pre-set message to their clinician to confirm they are ready for the next planned medication change when their average reading is above target.

| Patient and Public Involvement
Two PPI co-investigators were recruited at the outset as part of the core research team for BP:Together. Both had previously had a stroke and had experience of self-monitoring their BP. They helped write the protocol, attended all study team meetings and contributed to discussions regarding intervention procedures, content, qualitative study materials and data interpretation. One PPI co-investigator also recruited participants from her local Stroke Association group.
Additionally, a senior representative of Speakeasy (a charity supporting people with aphasia) helped to develop aphasia-friendly videos, information sheets and BP:Together booklets.
In the early planning stages (prior to studies 1, 2 and 3), two community groups also contributed to PPI activities (community

| Recruitment and procedure
Interviews and focus groups were conducted in phases with itera-

GP practices
Used search criteria to identify eligible patients based on history of stroke or TIA, and having a recent clinic reading of systolic BP above 130 mmHg. These patients received an invitation letter, a participant information sheet and a reply slip.

Community settings
To ensure inclusion of Black and minority ethnic (BAME) patients and those living in economically deprived areas, participants were also recruited from multiple community-level sources. TR built relationships with the leaders of local churches and mosques whose congregations included BAME communities, and presented the research during their prayer services. A formal recruitment request sent to communitybased staff from the Stroke Association, and attendance of two stroke support group meetings in economically deprived neighbourhoods in Oxfordshire and London further aided recruitment. PPI co-investigator MW visited the London group, enabling patient-to-patient recruitment.
Sampling of stroke patients was purposive based on age, gender, time since stroke/TIA and severity of post-stroke/TIA impairment in terms of understanding written information, speaking, vision, and arm and head movement. Participants reported these characteristics on reply slips returned to the team.
For study 1, face-to-face think-aloud interviews were conducted by TR in England and RD in Ireland. The interviews began by inductively exploring experiences of managing health after a stroke/TIA, and perceptions about BP. Participants then viewed the prototype BP:Together booklet, the website and/or the app with the researcher present, while saying out loud what they were thinking, thus sharing how they perceived it and potential barriers to engagement. 17 For study 2, emulating the upcoming randomized controlled trial (RCT), TR gave participants in-person training (Section 2.2) at their home and left them with a BP monitor, the BP:Together booklet and a diary to note any thoughts or feedback while self-monitoring at home (to help with their recall during the interview). Participants then used the intervention to self-monitor their BP at home over one week, and submitted their readings via SMS message, app or website. Retrospective, semi-structured telephone interviews then explored participants' experiences of using the intervention, specifically: self-monitoring their BP; submitting their readings using the BP:Together programme; and receiving feedback on their BP readings from the programme. Telephone interviews were offered because the researcher had already visited the participant during the training session and established rapport with them, and a telephone call could be both practical and convenient for gathering the data needed. Participants were offered the option of a second face-to-face visit for the interview if they preferred.
During development, the intervention was not 'live': communication regarding participants' BP readings was linked only to the study team and not to primary care.

| Study 3: Focus groups with health-care professionals
Invitations were sent to GP practices in the study area. Clinicians and support staff (GPs, nurse prescribers, practice or research nurses, administrative staff, health-care assistants) were invited by email (with an attached participant information sheet) from practices that expressed interest. We sought to gain insight into the process of implementing the intervention in primary care, from both the perspectives of clinicians and practice administrative staff who would provide practical support in processing the communication between patients and their GPs. Prior to the focus group, clinicians were asked to complete the online training module and record any thoughts to share. One or two facilitators were present at each focus group (KM, TR, CS or RD).

During the focus group, open-ended questions explored what partici-
pants liked and disliked about the intervention and training, and any facilitators or barriers to implementing it in practice.

| Data analysis
Using the PBA approach, 16,17 researchers moved cyclically between data collection, analysis (to identify potential changes to the intervention), modifications to the intervention and then further data collection. KM and TR familiarized themselves with the interview recordings and transcripts. Quotes from the transcripts were extracted and recorded in a 'table of changes' according to the relevant section of the intervention and whether they were positive or negative. This was a highly rigorous process, which involved rapidly collating all feedback relating to each aspect of the intervention, to facilitate discussion of potential iterative changes, drawing on evidence, theory, and PPI and expert opinion. Appropriate modifications were implemented by KM and CrR when deemed important to overcome barriers to engagement, thus making the intervention more acceptable, persuasive and feasible to implement. A set of criteria 34 were used to record the reason for and importance of making the change. The behavioural barriers identified from the qualitative data and PPI feedback, and the solutions used to address these in BP:Together are listed in Table 2. The four key recurrent barriers are discussed in more detail below, to demonstrate how optimization of the intervention was enhanced by seeking feedback from a diverse range of sources.

| Intervention materials for people with aphasia
Heeding early advice from our PPI co-investigators and community  There were also occasions where the stroke patient was less interested in learning about the intervention than other family members.

| Need for additional support
Habib, an elderly, Asian man who had recently suffered a second stroke lives with his wife and their son's family, and his daughter lives nearby.

TA B L E 2
The behavioural barriers that were identified from the qualitative data and PPI feedback, and the solutions used to address these in BP:Together Added an explanation to the BP:Together booklet that the intervention helps GPs to understand your blood pressure. Discussion at the baseline medication review will also help reassure patients that their GP is committed to the intervention.  The clinician online training was optimized to identify some of the common misconceptions patients hold about BP and stroke based on these insights into low motivation and health literacy, and trains clinicians to discuss these openly with patients at the start. Also, the BP:Together booklet was optimized to include a clearer explanation of the rationale for the intervention, emphasizing the value of controlling BP to avoid a subsequent stroke.

| Beliefs about changing medication
In a community PPI session, stroke patients and family members expressed preference for the medication change process to be initiated needed. 38 Working closely with the PPI co-investigators, the researchers carefully navigated their way through the stated reluctance from participants, trying out multiple iterations of the particular phrasing of the feedback message that invited them to contact the clinician following raised readings. For instance, asking patients to send a message saying 'I'm ready' rather than 'Change' (following an average reading recommending a medication increase) was an important distinction which patients were much happier with, as it avoided implying they were instructing the clinician, and instead simply communicated to their clinician that they were comfortable and ready to make the medication change.
The BP:Together booklet was also adjusted to clearly explain to patients the rationale for contacting the clinician about medication changes, including a fictional quote from a 'busy GP' explaining how receiving such a message helped her.
Clinicians in the focus groups felt reassured by patient involvement in the medication change process. Some felt that patients may not be willing to change their medication if it was purely clinician-driven.
You'd have to pick your patient. It's my experience that they don't take the tablets that you give them, never mind increase them.
(Health-care professional focus group 4) In response to these concerns, the baseline medication review in which the clinician makes the three-step plan with the patient was modified to include a guided discussion to manage both patients' and clinicians' expectations about medication change, reassuring each party that the other was happy with the process.

| Strengths and limitations
The main findings in this paper offer important insights that go

| Conclusions
In conclusion, thinking about how an intervention will work must begin by examining the factors that produce and sustain poor health in the first place. 14 Health problems that interventions seek to ameliorate should be explored in the social context in which they exist, 15 with an understanding of the mechanisms that render some people at greater risk than others. Any planned intervention must make clear how it anticipates responding to those problems. 14 Ultimately, the aim of public health practice is to improve health for all, especially those who have the greatest potential margin of benefit from the intervention. This work demonstrating the integration of the person-based approach and PPI will hopefully inform wider discussions regarding the development of complex health interventions to make them attractive, engaging, useable and most importantly beneficial, for the widest range of people.

ACK N OWLED G EM ENTS
We would like to thank all our study participants and PPI contributors for taking part in this study. We are grateful to the primary care practices that sent out participant invitation letters and helped recruit health-care practitioners for the focus group discussions. Thanks also to the following organizations for their help with participant recruitment and PPI input: the Stroke Association's London-based Support Groups;

Richard McManus has received BP monitors for research from
Omron and is working with them on the development of a telemonitoring system. There are no other conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. and recruitment to digital health interventions: a systematic review of qualitative studies. BMC Med Inform Decis Mak. 2016;16(1):120.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.
How to cite this article: Rai T, Morton K, Roman C, et al.