Shared decision making in consultations for hypertension: Qualitative study in general practice

Abstract Background Hypertension is mainly managed in primary care. Shared decision making is widely recommended as an approach to treatment decision making. However, no studies have investigated; in detail, what happens during primary care consultations for hypertension. Aim To understand patients’ and clinicians’ experience of shared decision making for hypertension in primary care, in order to propose how it might be better supported. Design Longitudinal qualitative study. Setting Five general practices in south‐west England. Method Interviews with a purposive sample of patients with hypertension, and with the health‐care practitioners they consulted, along with observations of clinical consultations, for up to 6 appointments. Interviews and consultations were audio‐recorded and observational field notes taken. Data were analysed thematically. Results Forty‐six interviews and 18 consultations were observed, with 11 patients and nine health‐care practitioners (five GPs, one pharmacist and three nurses). Little shared decision making was described by participants or observed. Often patients’ understanding of their hypertension was limited, and they were not aware there were treatment choices. Consultations provided few opportunities for patients and clinicians to reach a shared understanding of their treatment choices. Opportunities for patients to engage in choices were limited by structured consultations and the distribution of decisions across consultations. Conclusion For shared decision making to be better supported, consultations need to provide opportunities for patients to learn about their condition, to understand that there are treatment choices, and to discuss these choices with clinicians. Patient or Public Contribution A patient group contributed to the design of this study.


| INTRODUC TI ON
Shared decision making is a collaborative process through which a clinician supports a patient to reach decisions about their health-care treatments. 1 There are many models of shared decision making; 2-5 a recent influential model describes a process in which information is exchanged between clinician and patient about treatment options, and the patient's values and preferences, before choices are deliberated in collaboration. 5 The NHS long-term plan aspires to make shared decision making the usual experience for patients. 6 However, shared decision making has proven challenging to embed in routine care; [7][8][9] clinicians report time constraints and shared decision making not applying to all clinical situations. 10 Although providing information about treatment options can increase knowledge about those options, 11 this does not necessarily enable patients to participate in discussions or decisions about their care, and patients report feeling disempowered within consultations. 9 Challenges for shared decision making in UK primary care include that consultations often cover several problems, 12 diagnoses may be unclear, and the consultation may not focus on choosing appropriate biomedical treatment. 13 The distribution of decisions over time, courses of action, people and situations [14][15][16] might also present challenges for shared decision making. While some UK initiatives have shown promise in facilitating shared decision making, 8 few studies have focused on shared decision making for hypertension in the primary care setting.
Hypertension increases the risk of cardiovascular conditions such as strokes and heart attacks and is the leading preventable cause of premature death worldwide. 17 In the UK, hypertension affects 14% of adults 18 and is managed mostly in primary care, accounting for 12% of primary care consultations and approximately £1 billion in drug costs in 2006. 19 Recommended management options are based on age, blood pressure level and absolute risk of cardiovascular events and include drug treatments (antihypertensives and statins) and lifestyle modification. 20 Treatment is typically lifelong and adjusted over time. Optimal blood pressure targets vary internationally and by comorbidity and are the subject of vigorous debate. 21 Hypertension control has long been considered suboptimal; that is, it fails to reach specified treatment targets. 22,23 Blood pressure lowering reduces cardiovascular risk and delivers cost savings at the population level. 24,25 The likelihood of benefit / dis-benefit varies with cardiovascular risk and with the threshold at which hypertension is diagnosed and treatment considered. Shared decision making has the potential to ensure decisions are based on what matters to patients, informed by their assessment of the potential benefits and harms.
This study aimed to explore primary care patients' and clinicians' experiences of decision making for hypertension treatment, in order to understand if and how patient involvement in decision making might be supported.

| ME THODS
We chose a longitudinal design and qualitative methods to facilitate in-depth exploration of patients' and clinicians' experiences of decision making. Our aim was to follow patients as they attended health-care consultations and to understand the consultations both from their perspective and from the perspective of the clinician they had consulted. We did this using baseline interviews with patients at entry to the study, observations of consultations with a range of health-care practitioners and post-consultation interviews with patients and clinicians.

| Sampling and recruitment
We used purposeful sampling 26 to recruit practices varying in sociodemographic characteristics and organizational structures in a large southern English city. Within practices, we purposively sampled patients 18 years or older with hypertension to achieve a sample with a range of ages and at different stages after a diagnosis of hypertension. Adults with hypertension were identified using electronic record searches, mailed study information and asked to contact the research team if interested in the study. Sequential recruitment of patients and practices allowed sampling to be informed by early findings. All patients and health-care professionals provided written informed consent. The concept of 'information power' [24] was used to inform the patient sample size required. This meant data collection ended when the sample held sufficient information relevant to the study, to address its aims.

| Data collection
Data were collected between May 2017 and March 2018. All data were collected by RJ, a female clinician (general practitioner) and researcher, who was not involved at any stage in the clinical care of the patient participants or as a practitioner at the practices involved. One of the practitioners (GP, observed in 1 consultation) was known to RJ on a professional level. All included patients took part in a baseline interview. Patients informed the researcher about their health-care appointments, so that they could be observed and audio-recorded. Following consultations, health-care professionals and patients were interviewed separately to understand their perspectives on the consultation. Brief topic guides were used for both baseline and post-consultation interviews. Patient participants were followed for up to six health-care appointments to allow observation of experiences over time. Field notes recorded observations about the health-care setting and consultations and served as a prompt for post-consultation interviews. Baseline and post-consultation interviews were audio-recorded, and, with a small number of exceptions (due to researcher unavailability or technical failure of recording equipment), consultations were both observed and audio-recorded.

| Data management and analysis
Interview and consultation data were audio-recorded, transcribed verbatim and anonymized. NVivo was used to aid data management and analysis. Data analysis was thematic, 27 used an inductive, constant comparison method and was concurrent with data collection.
Three authors (RJ, HC and KT) developed and refined the coding frame. KT and HC are qualitative methodologists: KT has a disciplinary background in social sciences applied to health; HC has a background in social and medical anthropology.
Data for each participant were analysed as a case. Data were explored for similarities and differences within patient case studies (between patients and clinicians, between consultations and post-consultation interviews), over time, and between case studies. Codes were built into broader categories and themes; themes were developed inductively from analysis across the case studies.
As data collection continued, we re-assessed the factors influencing information power 28 and ended data collection when we judged sufficient information power had been achieved.
We chose not to use a specific theory of shared decision making to inform the analysis, as we were keen for our findings to be inductively developed from the data. Also, a systematic review of shared decision-making models found little agreement on the concept, identifying only two features (patients values / preferences, and options) appearing in more than half of the models. 2 We therefore adopted a broad understanding of shared decision making, that is decision making in which both patients and healthcare professionals are involved, while sensitizing our analysis to the presentation of options and discussion of patients' values and preferences.
The study Patient and Public Involvement group comprising seven members (four patients with cardiovascular disease and three of their partners / carers) assisted in development of the study materials and in the submission of an ethics application, which was reviewed and approved by the NHS ethics committee (Research Ethics Committee reference: 16/SW/0294REF).

| Participants
Five general practices in Bristol, England, were recruited. Eleven patients with hypertension were recruited; participant characteristics and data collection are summarized in Table 1

| Poor understanding of hypertension and its treatment
During the interviews, patients reported understanding that having high blood pressure meant they might have a heart attack or stroke ( Figure 1). However, understanding of high blood pressure and cardiovascular risk was often limited. Patients reported lack of time and discomfort at asking questions as reasons they did not ask ques-

| Distributed decisions
Decisions about treatment were often distributed, that is, as an ongoing event evolving across encounters and over a range of people ( Figure 2). 15 They were also sometimes revisited, over time and  Case study: Ray Despite a series of appointments with his GP, Ray did not develop their understanding of their blood pressure over the course of the study. Ray was anxious that his blood pressure might have caused heart damage and, felt at risk of having a stroke, understanding high blood pressure as a 'silent killer'. Ray was concerned to get his blood pressure down, but did not understand what blood pressure level to aim for; this was never discussed in his consultaƟons, and despite searching on the internet between his consultaƟons, he was unable to be sure. Ray did not discuss his worries with his doctor. Throughout the study, the GP took Ray's home blood pressure measurements into account when making decisions about treatment, despite there being no discussion about how Ray was monitoring his blood pressure, or the goals of treatment. In Ray's fourth observed consultaƟon, the failure of Ray's blood pressure to respond to treatment led to a brief discussion in which it was established that Ray was using a potenƟally inaccurate (wrist) blood pressure monitor. This informaƟon was crucial in the GP's assessment that he may not need addiƟonal medicaƟons. In this example, the lack of discussion of how Ray was measuring his blood pressure had led to decisions being made on the basis of potenƟally inaccurate figures over the course of several consultaƟons, and opportuniƟes to build Ray's understanding and confidence were not taken.

Distributed decision making
Case study: MaƩhew In his baseline interview Mathew described seeing his GP because of headaches three years ago, leading to a series of appointments about his blood pressure: 'one of the things she did was test my blood pressure and said its quite high…it was a concern so we didn't do anything about it at the Ɵme, then we went back again and they tested it again and said no its sƟll high so they fiƩed me with one of those-I had one of those 24 hours, you know, things that just tested my-tested me all the Ɵme and they did come back and go it's a bit high but it wasn't-it wasn't ridiculous that they were suddenly trying to, you know, put me on pills and I'm not a great fan of taking medicaƟon if I don't need to so I went back and talked to my wife who's a bit more of a health freak and she put-made me drink lots more beetroot and various bits and pieces … I got my own [ Case study: Adrian Adrian's blood pressure treatment decisions were distributed over a series of consultaƟons with his GP (four observed consultaƟons and one telephone consultaƟon) and a pracƟce nurse (one annual review). Each of the observed consultaƟons in the surgery was about several healthcare problems, and only a small amount of Ɵme was spent discussing his blood pressure. Adrian provided home blood readings at his nurse blood pressure review, and the nurse informed the GP. The GP advised Adrian, during a phone appointment to take a medicaƟon for his blood pressure. Adrian, aŌer reading the informaƟon leaflet describing potenƟal side-effects, decided not to take the medicaƟon and reported this to his GP, who reassured him and encouraged him to try the medicaƟon. A further GP appointment was made, and again changes to medicaƟon were suggested by his GP. In a post-consultaƟon interview, Adrian's GP acknowledged that he was assessing Adrian's blood pressure over several consultaƟons (establishing the severity of his hypertension through a series of blood pressure measurements and other tests). Over the course of several consultaƟons, Adrian's experience of side-effects was discussed, Adrian asked about the mechanism of acƟon of the medicaƟon, and invesƟgaƟons were completed. Despite having some opportuniƟes, over Ɵme, to ask quesƟons, opportuniƟes were brief and informaƟon received piecemeal. In post-consultaƟon interviews it was clear that Adrian did not understand the answers he had received, or the raƟonale for treatment changes; throughout the study he was not presented with opƟons.  June Ok, bye.
AŌer the consultaƟon Joan described her disappointment, saying that, for her, having informaƟon was the most important thing. She did not interrupt the nurse to ask for informaƟon, as 'she (the nurse) was on a roll', and thought that the nurse 'wasn't supposed to discuss much with her'.
Fieldnotes recorded aŌer Joan's consultaƟon: Very efficient consultaƟon (lasted 10 mins, she later told me they have 20 mins). Nurse spent a lot of the Ɵme looking at the computer (filling template), or direcƟng her gaze to a parƟcular taskchecking blood pressure, looking at the scales, washing her hands etc. LiƩle eye contact Recorded BP, during which paƟent broke the silence to comment 'probably quite high'; no response at all from nurse.

DeliberaƟng treatment choices in a consultaƟon
Opening a consultaƟon with a treatment request Alice opened her consultaƟon with a request:

Alice I'm sƟll worried about the Amlodipine and I was wondering if I can come off that completely 'cos I'm geƫng a lot of palpitaƟons, feeling Ɵred all the Ɵme and weak. Dr
Let's just take a step back. Alice Right. I'm calming down now. (laughs) Dr I just went quickly, skimmed through your blood pressure history which has been-Both the blood pressure and the treatment has been up and down hasn't it over the years-Alice I know Clinician's weighing up complex informaƟon In a post-consultaƟon interview, Ivy's clinician (clinical pharmacist) explained what he was weighing up in considering her medicaƟons:

I suppose from my point of view she's coming in wanƟng to stop a staƟn which probably doesn't maƩer too much but actually I think she might as well stay on it, she's not got problems with it, so I think it's good to encourage it although she's 81 so does she really, you know, the evidence is ten years isn't it, you start on staƟns-So that seemed reasonable…and then she's also-she has got terrible problems with her legs at the moment and I know she's seen lots of people with her legs so she's probably quite stressed as well so-But she's got CKD3 as well so probably we should be driving it a bit lower. (clinical pharmacist, interview)
Experiences of involvement: Joan, reflecƟng on who made the decision to increase her blood pressure medicaƟon: It was his decision. Yes, and I did make a liƩle protest about the fact that it was high when they took it in that consultaƟon when I had my annual check-up but he said he thought the week's readings that the average of that was a bit high. So I went along with him .
Adrian's GP reflected on decisions he had made with Adrian: Interviewer Ok. Would you describe those decisions as decisions you shared with him? Doctor I think-Yes but I think they were iniƟated by me obviously but I think he was quite happy to go along with-I think I was probably leading him on more than anything but I think he was happy with the explanaƟons.
Case study: Matthew For one paƟent, consulƟng with a clinical pharmacist, medicaƟon choices were considered at length. MaƩhew opened the consultaƟon by establishing that he had quesƟons to discuss, aŌer which he and the pharmacist engaged in a detailed discussion about MaƩhew's home blood pressure monitoring, including explicit discussion of blood pressure treatment goals and sideeffects, before the pharmacist proposed a treatment change with a detailed raƟonale referencing naƟonal guidelines and explicitly seeking MaƩhew's views on the proposal, responding to MaƩhew's quesƟons about the 'pros and cons' of treatment before a decision was reached. between different consultations and different health-care professionals. Most decisions were made by GPs, but the processes of decision making, notably collecting information relevant to a decision, involved other health-care professionals, for example nurses.
Sometimes information relevant to a decision was collected on one occasion, but informed a treatment discussion on another occa-

| The constraints of structured consultations
Consultations with a range of health-care practitioners were observed to be highly structured and led by the health-care practitioner ( Figure 3). The consultation structure typically began with a greeting, after which the practitioner asked a series of questions to collect information, followed by a physical assessment. Observed consultations with nurses included routine yearly reviews. During these, data collection focused on lifestyle behaviours and examination included measurements such as blood pressure and weight. Asked after a consultation what her role was, one nurse responded 'just to literally -go through the template, just ask them about the diet and exercise, just to inform them what I'm doing' (nurse, interview). Brief lifestyle advice was given in response to behaviours reported, and deviations from the routine were infrequent and brief. In consultations with GPs, information gathering included symptoms, medication taking, side-effects and lifestyle behaviours. Physical assessment was often followed by the GP reviewing medication lists and recent investigations, before reporting their findings and making a plan.
Consultation routines were practitioner-led, with patients providing responses to the practitioner's questions. Routines appeared to constrain the opportunities for the patient to contribute to the consultation. As Ivy commented, 'consultations are about one specific thing and the doctor pulls you back to these' (patient, interview). At the beginning of the consultation, some clinicians opened the consultation with a question asking the patient what they wanted to address. David's GP opened the consultation by asking 'what do we want to start with?'(clinician, observation). Typically, later questions to the patient were seeking agreement to the clinician's suggestions, or checking that the patient had no further questions.
Patients reported that the structured nature of the consultation was frustrating.

| Limited involvement in deliberating treatment choices
During observed consultations, changes to treatment were usually suggested by health-care professionals (pharmacists and doctors) after they had completed their assessment (Figure 4). Treatment proposals were sometimes preceded by a brief summary of the assessment and rationale for the suggested change. Occasionally patients opened the consultation with a proposed treatment change, sometimes triggered by the experience of side-effects; when they did so, this often led to a change in medication. Clinicians typically suggested one treatment without alternatives, or occasionally more than one option (eg to increase medication or pursue lifestyle changes). When GPs wanted to increase a patient's hypertension medication, in no instances did they offer the option of not increasing it, or offer alternative anti-hypertensive medications. Reflecting on a decision to increase her blood pressure medication Joan said : 'He didn't suggest anything else', and when asked if side-effects were discussed she answered 'No. Didn't think about that' (patient, interview). There was no detailed discussion of risks and benefits of treatments, and in only one observed consultation was the patient asked for their perspective on what they wanted to achieve, their treatment preferences, the proposed treatment or alternative treatment options. Observations and postconsultation interviews with clinicians indicated that, often, many decisions were being deliberated by clinicians as they assessed the patient, but these deliberations were not made explicit or shared with patients.
During the interviews, many patients reported satisfaction with their involvement in decision making. John said 'you want the best decision for yourself really, that's why-she tells you to take they tablets, you're putting your hand in hers to try and sort it out 'cos she's a doctor' (patient, interview). For some, agreeing to a treatment or participating in a brief discussion about treatment was perceived as involvement in the decision. However, some patients reported being dissatisfied with decision making in consultations. For example, Adrian expressed frustration at being only offered increasing amounts of one medication, over a series of consultations, to lower his blood pressure, saying 'You never really have enough time to ask other questions and get another solution. Is there any other thing I could take apart from Amlodipine, he's already chosen the best for me I suppose and you've got to trust the doctor'. Alice, who had developed new health problems attributed to a blood pressure medication that she had insisted remaining on, was angry that her GP had appeared to accept her request without discussion. Neither Adrian nor Alice discussed this with their GP.
Many doctors felt that patients would be able to voice their disagreement with the doctor, if needed. When side-effects were experienced, some GPs reflected that the right course of action (to change the treatment) was evident, while for treatments that aimed to prevent future events rather than manage a symptom, GPs were more likely to lead the decision. One clinician described how some choices were for the patient to make (whether to persist with lifestyle choices or accept increase anti-hypertensive medication), but if blood pressure was persistently above target after

| Summary
Little shared decision making was described by patients or ob-

| Strengths and limitations
The use of multiple qualitative methods and a longitudinal design allowed us to explore decision making from multiple perspectives, to understand how, when and why decision making did or did not occur, and to do so over time. It is accepted that the presence of a clinically trained observer in consultations might influence the behaviour of practitioners and patients. 29 Participants were aware that the study

| Comparison with existing literature
Few studies in primary care have used observational methods to explore shared decision making and we are not aware of any focusing on hypertension decision making. While shared decision making is endorsed widely in national guidance, 30,31 robust evidence to guide a choice of intervention to support shared decision making for hypertension, or to understand is health outcome, is lacking. 32 In this study, as in others, little shared decision making was observed, 33 yet some patients were satisfied with their limited role in treatment decisions. This may reflect a true preference not to be involved.
Alternately, it may reflect an internalized set of normative expectations about their participation, 34 or the desire to be a 'good patient' and avoid conflict within the encounter. 9 Elwyn et al's collaborative Qualitative studies in different contexts report that patients prefer collaboration, including sharing concerns and receiving explanations, but many feel they lack sufficient expertise, or power, to make decisions, 9,35 and often clinicians do not explicitly mention or provide detailed information about treatment options. 36,37  Clinicians follow a structured approach to consultations, formalized in consultation models, one aim of which is to assist the clinician in negotiating the complexity of the interaction between patients and doctors. 42 Prominent models establish the consultation as task-focused, while encouraging patient-centred elements such as exploring the patient's ideas, concerns and expectations. 43,44 In this study, consultation routines could be discerned and appeared to be used to manage consultations efficiently by ensuring that biomedical tasks were addressed, but were infrequently used to focus on probing patient concerns. Opening up space in consultations to better understand patient perspectives on treatment may be necessary for shared decision making, but may have consequences for the efficiency of consultations to achieve other important tasks. In this study, we described distributed decision making [14][15][16] which made it more difficult to identify decisions and resulted in a lack of transparency and the exclusion of patients from the decision-making space. 16

| Implications for research and clinical practice
The provision of universal personalized care, including shared decision making, is a central tenet of the NHS 10-year plan. 6 For shared decision making to happen, treatment options must be identified and made explicit. More work is needed to clarify which decisions should be shared between doctor and patient, and how best to do this. Future work could analyse consultations for hypertension using validated shared decision-making tools. As a minimum, patients should be aware when decisions are being made about them, and should understand how the information they share with clinicians is taken into account when decisions are made. For shared decision making to be supported, patients need better understanding of their hypertension, and their understanding and treatment preferences need to be explicitly sought. Shared decision making is not possible when patients and clinicians do not have a shared understanding of the decision they are facing and patients do not understand what is at stake when decisions are made.
Efforts to support shared decision making need to focus on how consultations are organized, and how decisions arise within, and across consultations. It is important to identify the potential ways in which shared decision making could be supported in the context of distributed decision making, which is likely to be commonly encountered in primary care. For example, distributed decision making might provide multiple opportunities for patients to develop informed preferences, taking into account perspectives from a range of health-care providers, provided the decisions being considered are made explicit, and health-care professionals have the required skills. 14 Consultations need to achieve biomedical tasks alongside addressing the patient's agenda. Given the inherent power imbalance in the doctor-patient relationship, 9 this may require re-structuring consultations, in order to make patients more equal partners in the conversation.

E TH I C S A PPROVA L
This study was reviewed and approved by the NHS ethics committee (Research Ethics Committee reference: 16/SW/0294REF).

Many thanks to Professor Nicky Britten, Professor Elizabeth
Murray, Professor Richard Lehman and Professor Glyn Elwyn for their formative advice during the conceit of this study and to Professor Adrian Edwards for helpful comments on the draft paper.
Many thanks to the PPI group who helped throughout this study.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request due to privacy/ethical restrictions: 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.