‘I should have taken that further’ – missed opportunities during cardiovascular risk assessment in patients with psoriasis in UK primary care settings: a mixed‐methods study

Abstract Background Unhealthy lifestyle is common in psoriasis, contributing to worsening disease and increased cardiovascular disease (CVD) risk. CVD risk communication should improve patients’ understanding of risk and risk‐reducing behaviours; however, the effectiveness of risk screening is debated and evaluation currently limited. Objective To examine the process of assessing for and communicating about CVD risk in the context of psoriasis. Design Mixed‐methods study in English general practices to (i) determine proportions of CVD risk factors among patients with psoriasis at risk assessment and (ii) examine patient and practitioner experiences of risk communication to identify salient ‘process’ issues. Audio recordings of consultations informed in‐depth interviews with patients and practitioners using tape‐assisted recall, analysed with framework analysis. Participants Patients with psoriasis (n = 287) undergoing CVD risk assessment; 29 patients and 12 practitioners interviewed. Results A high proportion of patients had risk factor levels apparent at risk assessment above NICE recommendations: very high waist circumference (52%), obesity (35%), raised blood pressure (29%), smoking (18%) and excess alcohol consumption (18%). There was little evidence of personalized discussion about CVD risk and behaviour change support in consultations. Professionals reported a lack of training in behaviour change, while patients wanted to discuss CVD risk/risk reduction and believed practitioners to be influential in supporting lifestyle management. Conclusions Despite high levels of risk factors identified, opportunities may be missed in consultations to support patients with psoriasis to understand CVD risk/risk reduction. Practitioners need training in behaviour change techniques to capitalize on ‘teachable moments’ and increase the effectiveness of risk screening.


Introduction
In the UK, the National Health Service (NHS) health check programme 1 was established in 2009 to reduce the incidence of cardiovascular disease (CVD) and associated mortality, as well as inequalities in health. 2 The programme offers CVD risk assessment to all individuals aged 40-74 years without existing CVD, diabetes or metabolic conditions, consisting of physiological measurements such as blood pressure (BP)/ cholesterol as well as profiling of medical family history and lifestyle factors such as smoking. Guidelines state that individuals identified as having an elevated risk should be appropriately supported with lifestyle modification and/or pharmacological intervention to reduce such risk. 3 The health check programme debate Health check attendance levels and treatment uptake have been lower than anticipated 4 and evaluation in one large English health region found the checks failed to identify one-third of people at high risk of developing diabetes. 2 Krogsbøll et al.'s international Cochrane review 5 suggests that health checks reduce neither morbidity nor mortality, but do increase the number of new diagnoses and pharmacological interventions, concluding that their benefits do not outweigh the associated harms. Others have also shown little change in reported prevalence of morbidity resulting from health checks. 6 Subsequent debate among clinicians, researchers and policymakers about the effectiveness of health checks at the population level 7 has led to calls to abandon them. 8 Disadvantages of the programme include the following: low uptake, 9 provision of inappropriate pharmacological solutions and false reassurance, 10,11 as well as discounting behavioural risk reduction strategies. 12 However, the Cochrane review has been criticized for poor methodological quality, 13 and other commentators stress the positive improvement in appropriate prescribing of statins following health checks. 14,15 Studies to evaluate the programme have so far been limited 9,16 and qualitative approaches to explore patient and practitioner experiences of health checks should form part of future evaluations. 17

Psoriasis and CVD risk
Psoriasis is a complex, long-term inflammatory systemic condition, presenting with a skin rash, affecting at least 2% of the UK population. 18  People with psoriasis often live long-term 19 with a range of challenging demands including high levels of chronic physical and psychosocial disability. 20,21 While severe psoriasis is associated with CVD, 22,23 the precise nature of this relationship is not yet fully understood. 24 Unhealthy lifestyle behaviours such as excess alcohol consumption, 25,26 smoking, 27,28 inactivity and higher BMI 29 are all known to be more common in people with psoriasis than the general population. These behaviours may have a role both in psoriasis onset and exacerbation or severity, as well as being risk factors for CVD. 28,[30][31][32][33][34][35] High levels of psychological distress associated with psoriasis 36,37 can increase risk behaviour and reduce motivation or capacity to engage in healthy behaviours. 38 The National Institute for Health and Care Excellence (NICE) guideline for England and Wales on assessment and management of psoriasis 39 recommends that practitioners identify and address comorbidities with patients, including discussion of lifestyle management.

CVD risk communication and risk reduction
Perception of personal risk is highly complex and prone to systematic or error-based biases. 40,41 The ways in which individuals perceive risk are influenced by their prior awareness and understanding of the risk as well as how it is presented to them. 42 Communicating CVD risk is particularly complex because risk calculation/ assessment involves amalgamating information about a range of physiological and lifestyle factors to predict patients' future risk, often in the absence of current symptoms. Health professionals may use population risk values which lack personal relevance to patients, 43 whereas personalized risk information may be required to increase risk understanding. 44 Furthermore, the effect of communicating CVD risk upon the likelihood of engaging in risk reduction behaviours is unclear.
Qualitative research with healthcare practitioners shows that their personal attitudes towards/beliefs about health checks can be barriers to delivering lifestyle change support to patients 45 and that practitioners' skill sets could be enhanced by becoming confident and competent in the use of behaviour change techniques. 46 Similarly, the literature suggests that patient barriers to engaging with and maintaining lifestyle modification include beliefs, emotions, information needs and social support. 47,48 This study aimed to explore patient and practitioner perspectives of the process of assessing for and communicating about CVD risk, using psoriasis as an exemplar of a condition where risk reduction in the form of lifestyle modification may be beneficial to disease management.

Methods
The study involved two elements: (i) CVD risk factor assessment of adult patients with psoriasis above and below 40 years of age (to include younger patients not usually invited for a health check) in 13 general practices in North West England; and (ii) a 'nested' qualitative study of the process of risk communication in consultations for CVD risk assessment in people with psoriasis through in-depth interviews with patients and practitioners, using excerpts from audio-recorded risk communication consultations to assist questioning. People with psoriasis are not specifically targeted for CVD risk assessment; however, general practitioners (GPs) and practice nurses were asked to conduct risk assessment consultations in accordance with their usual health check procedures to capture routine practice.

Sampling and recruitment
General practices varying in size and locality were identified and recruited from five primary care trusts across North West England. Practices were reimbursed for their participation. Using Read codes known to map to psoriasis and medications/topical preparations for psoriasis, practices identified patients with psoriasis over 18 years old on their lists. Those with severe mental health problems, without capacity to consent, the recently bereaved and the terminally ill were excluded by the GP. Identified patients were invited to attend a CVD risk assessment at their practice. Smaller practices mailed all eligible patients on their list; larger practices mailed an agreed number (depending on practice capacity), in which case a random number list was used to select potential participants.
Ethics approval was obtained from the North-West Research Committee, Greater Manchester East (REC ref: 11/NW/0654). Patients expressing interest were telephoned by a researcher who explained the study and, with the patient's agreement, arranged a CVD risk assessment at their own practice. Consent to audio-record the risk assessment and follow-up consultations was sought from all participating practitioners and patients, with the aim of capturing as many recordings as possible across the sample. Participants whose consultations were recorded were sampled to undergo subsequent qualitative interviews on the basis of (i) consultations in which salient process issues were identified and (ii) diversity of personal characteristics. Practitioners were sampled to obtain a mix of GPs/ practice nurses and patients for diversity on age and gender.
Quantitative data (CVD risk assessment)

Data collection
On attendance at the practice, informed consent was acquired from patients and each completed a medical history questionnaire. The GP or practice nurse recorded a range of biomedical and behavioural measurements that are part of standard CVD risk screening procedures (see Table S1 for measurements recorded at risk assessment). Patients were recalled for a followup discussion of their screening results if deemed necessary by the practice. The research team was informed if a follow-up appointment was advised.

Data analysis
Risk factor variables were derived from measured or patient-reported data based on current published UK guidelines. 49,50 Qualitative data (CVD risk assessment processes)

Data collection: audio recordings and interviews
Consultations were audio-recorded and used to inform subsequent in-depth interviews with practitioners and patients about their experiences of risk communication during consultations using 'tape-assisted recall' (T-A-R). 51 The T-A-R approach enabled the interviewer to probe participants' reflections through replaying excerpts from consultations to ground questioning in specific examples of risk communication or discussion of lifestyle management. Interviews with participants were conducted between August 2012 and November 2013 and guided by interview topic guides (see Table 1 for content of the interview schedules). Interviews were audiorecorded, transcribed verbatim, anonymized and transferred to NVivo 10 for data management (QSR International Pty Ltd. Version 10, 2012). 52 Data analysis: audio recordings and interviews Two authors (AC and CP) familiarized themselves with the audio recordings to identify process issues (when and how CVD risk was discussed/addressed in the consultations) using a qualitative content analysis approach. 53 An a priori analysis framework of topics relevant to CVD risk assessment guided critical listening/ coding of audio recordings (see Table S2). Coding enabled identification of instances when biomedical and behavioural factors as well as additional factors pertinent to this specific patient group such as psoriasis severity, quality of life and mood were discussed. Additionally, by coding the audio recordings, practitioners' communication techniques could be categorized in terms of (i) acknowledgement of patient cues for discussion, (ii) general communication style and (iii) approaches to addressing issues connected to risk/lifestyle. The two authors each analysed 50% of the audio recordings and discussed all coding to resolve and agree codes for ambiguous examples. The content analysis enabled identification of how CVD risk was addressed within consultations to guide selection of T-A-R excerpts to inform subsequent qualitative interviews. Principles of framework analysis, 54 developed for use in applied healthcare settings, were used to analyse the interview data. This approach facilitates investigation of pre-set as well as emergent topics using constant comparison 55 as data collection and analysis are conducted simultaneously until data saturation is achieved and no new insights are being generated. The research team met regularly to discuss key ideas emerging within and across interviews with particular attention to cases that differed from salient trends in the data and jointly agree a framework of main themes. Data were coded in NVivo enabling extraction of illustrative data.

CVD risk assessment study
Thirteen general practices participated. Practice size (number of registered patients over age 18) varied between 1086 and 16 746 patients. Practices were located in a variety of rural and urban areas and varied in levels of deprivation. From 1446 invitations for CVD risk assessment sent, 447 people responded and 287 attended an appointment (220 aged 40 and over; 67 aged under 40), an overall attendance rate of 20%. Of those attending, 165 (57%) were female and the mean age was 53 years. Comparing those who did and did not attend, there was no difference in the proportion of women; however, a lower rate of attendance was recorded for the under 40s (15% compared with 25%, respectively).
Biomedical data were analysed for proportions of risk factors in the sample (see Table 2 for risk factor definitions) evident at risk assessment and thereby amenable to discussion during the consultation. Table 2 reports proportions of the sample with the following CVD risk factors at risk assessment: selfreported smoking, self-reported alcohol consumption, obesity, very high waist measurement and raised BP. The most common risk factors seen at the risk assessment were obesity (35%) and a very high waist measurement (52%) indicating central adiposity. At risk assessment, raised BP was found in 29% of those attending. This figure includes patients already prescribed antihypertensive medication (but not reaching treatment target) as well as those with previously unknown hypertension. Eighteen percentage of those assessed reported smoking and 18% reported drinking above the recommended units of alcohol per week. Audio-recorded consultations Practitioners in 10 of the 13 participating practices agreed to consultation audio recording. A total of 130 CVD risk assessment and 15 follow-up consultations were audio-recorded (involving four GPs, nine practice nurses and 131 patients with psoriasis). Risk assessments were generally conducted in dedicated clinics making audio recording straightforward; however, follow-up consultations were conducted on an ad hoc basis according to the practice's usual follow-up procedure. As a consequence, the practices experienced major difficulties capturing follow-up audio recordings. Analysis of the audio-recorded consultations revealed three core issues during patientpractitioner interactions. Firstly, there was little evidence of detailed discussions about CVD risk (core issue 1). When patients offered cues about concerns related to risk factors (e.g. consultation number 111: patient with high BP cues '. . .it's why I'm on this health kickto get the weight down'), practitioners commonly responded by blocking or failing to pursue patients' concerns, shutting down rather than opening up a dialogue. Secondly, practitioners demonstrated a focus on recording information rather than opportunistically addressing CVD risk reduction in consultations (core issue 2). This was evident in instances where a clear statement from the patient (e.g. consultation 101: 'I'd love to lose a bit of weight to be honest') elicited a disconnected practitioner response. Here the clinician appeared to be primarily fixed on documenting biomedical or behavioural data rather than incorporating the patient's agenda into the work of the consultation. Thirdly, little evidence of skilled patient-centred practice was observed in consultations (core issue 3). Skilled practice was apparent in a minority (e.g. consultation 29: practice nurse, having determined the patient was not yet ready to stop smoking, closed the consultation with a prompt towards the possibility of future behaviour change '. . .so think about your smoking. . .you can either come to the GP or the smoking cessation clinic here, they'll accommodate you if you want some support with that'). This was a rare example of a practitioner recognizing and acting upon an opportunity to increase the salience of possible behaviour change in the patient's mind and support their self-efficacy. This practice nurse was one of only two practitioners (both practice nurses) who consistently picked up on and responded to openings in consultations to address CVD risk with patients.
In summary, opportunities to discuss and address lifestyle behaviours such as smoking, diet and exercise which occur naturally in risk assessment consultations were often overlooked by practitioners in these recordings. Further extracts from the audio-recorded consultations are presented in Table 3 as illustrative examples of instances where practitioners appeared to miss or use opportunities to address patient cues for discussion of CVD risk and/or lifestyle management.  Nurse responds directly to patient's concern, signals that she will support patient, explores patient's confidence to make a change 4,5 Gives rational for increasing/maintaining physical activity, encourages individualized choice 4,5 Moves towards planning for change 5

61)
PN9: But first of all just as a first thought.

Patient and practitioner interviews
In-depth interviews were carried out with 29 patients (18 female; 12 < 40 years of age) and 12 practitioners (10 female; eight practice nurses and four GPs) who had consented to have their consultations audio-recorded (see Table S3 for characteristics of interviewed patients and practitioners). Interviews focused on patient and practitioner perceptions of CVD risk assessment/ follow-up consultations. Excerpts from the audio-recorded consultations, including examples of instances when opportunities for discussion of CVD risk and/or risk reduction appeared to have been missed, informed questioning.
Three key themes emerged from the analysis of the interviews: (i) limited shared discussion about CVD risk and lifestyle issues, (ii) limited provision of personalized risk reduction support to patients and (iii) the perceived influence of health practitioners in supporting risk reduction. These themes expand upon and illuminate the core issues (1, 2 and 3) revealed by the consultation data. Each is described below alongside relevant data extracts which highlight key contrasts in patients' and practitioners' views about the process of CVD risk assessment.

Theme 1: Limited shared discussion between patients and practitioners about CVD risk and lifestyle
Patients assumed that it was not their place to raise concerns in CVD risk assessment consultations because the process was essentially clinician-rather than patient-driven. This led them to minimize the importance of risk and lifestyle-related issues and contributed to the lack of discussion (core issue 1) apparent in consultations: Furthermore, patients assumed that practitioners would view these issues as trivial and not a good use of their time. Some patients reported being too anxious to be able to raise health concerns with practitioners in the consultation and others seemed keen to preserve relationships with clinicians by not bothering them with 'unimportant' concerns: Practitioners reported that their main focus in these consultations was on gathering information from patients and informing/educating them about healthy lifestyle behaviours with reference to government recommendations or advice. This suggests that they were less focused on patient-led discussion in which they could use cues in consultations to talk about patients' own concerns and explains the focus on information recording and lack of detailed discussion apparent in the consultations (core issues 1 and 2): identified were as follows: (i) The target group was atypical (adult patients with psoriasis of all ages); (ii) more time was allocated for consultations than in normal practice; and (iii) the consultation needed to be audio-recorded: It's not very different from doing a blood pressure clinic in terms of questions we need to ask, information we need to find out from patients, and then the advice that we give to them, so from that perspective it wasn't a big challenge. It's just the fact that things like the tape recorder, I think that puts a different slant on it. PN1 I saw it as more like a screening process really. . . like doing a well person's check, but the patient has psoriasis really. I do a lot of reviews anyway, like diabetic and heart reviews. So. . .to me it was a screening thing. PN2 Taken together, patients' reluctance to instigate conversations with practitioners about risk-related health concerns and practitioners' focus on information gathering/advice giving meant that there was little shared discussion taking place during consultations that enabled patients to understand their risk status or ways to reduce risk. Practitioners reported that the conduct of the consultations differed little from that of customary health checks, suggesting that shared discussion may also lack prominence in routine practice. There was limited evidence from practitioners' accounts of tailored support being provided in response to specific information provided by individual patients, echoing the low level of skilled patient-centred practice evident in consultations (core issue 3). It was notable that many practitioners expressed the view that their patients already possessed the knowledge required to understand and act appropriately to reduce CVD risk: Most people know what they should and shouldn't be drinking, eating, exercisingshouldn't they? PN4 I think most people know where they're going wrong, they know that they shouldn't be eating three cakes every day. . .it's in the media all the timeget five a day in, take more exercise, eat less, move morepeople know. PN7 Even when patients had several risk factors, they appeared not to prioritize discussion of them in the absence of current symptoms, with some seeing a need for action only once disease with perceptible symptoms had developed: You tend to think that unless there's actually something wrong, they don't call you in. Well that's always been my belief. P8: Male, aged 54, obese, very high waist circumference, 24 units weekly alcohol Until I develop cardiovascular disease, I'm not going to take much interest. . .if it happens I'll deal with it then. P3: Female, aged 64, obese, very high waist circumference Some interpreted the lack of directive discussion by the practitioner as a signal not to change anything about their lifestyle, believing that the messages did not apply to them. This lack of individualized risk/lifestyle discussion in consultations (core issues 1 and 3) could lead patients to miss the personal relevance of health messages about CVD risk: their role as providing standardized advice during CVD risk assessments, but at the same time, assumed patients already possessed the information required to make lifestyle changes. In contrast, patients often interpreted the absence of personalized engagement or advice from practitioners as a signal that the information was not relevant to or directed at them specifically, the implication being that they did not need to attend to risky health behaviours. This highlights the unintended consequences of the lack of detailed discussion and skilled patient-centred practice found in the consultations (core issues 1 and 3). Patients viewed primary care practitioners as approachable and trustworthy individuals with whom to appropriately discuss potentially sensitive concerns connected to body image or lifestyle behaviours, and perceived them to be in a position of influence in prompting their thinking on these issues: I'd rather talk to a health practitioner than somebody at the gym who's stick thin, loads of muscles and a bit intimidating. So a nice normal average person who's got the information is a lot more In contrast to this patient perspective, practitioners (generally practice nurses, who, more than GPs, were tasked with risk assessment/ reduction activity) reported views of themselves as lacking the power to effectively support patients with lifestyle changes. Nurses' accounts highlighted a lack of confidence in broaching potentially sensitive behaviour change topics in case the patient-practitioner relationship was disrupted. In addition, they expressed pessimism about their degree of influence to motivate patients: With hindsight I should have definitely taken that further [discussion with female patient drinking 30 units of alcohol per week]. I find that really interesting for reflective practice. . .I am too people pleasing. PN6 If they tell you they enjoy smoking I feel there's no point. They can't see the damage that's being done so they feel fine. You get 'my granddad lived to be 102 and he smoked'. I can't say anything about that. . .you know that you're never going to get anywhere with them, so that's up to them. PN3 Furthermore, nurses suggested they were not well equipped to do such work, being unsure how to check patients' understanding of risk/ lifestyle issues. Only one nurse (PN9) had undergone specific, structured training in techniques to support lifestyle behaviour change. The remainder reported that a lack of training undermined their ability to undertake these tasks which they simultaneously identified as being key elements of their professional role. This offers insight into the lack of detailed discussion and skilled patient-centred practice observed in the consultations (core issues 1 and 3): Patients viewed practitioners as being in a position to motivate them, to help them understand and address their CVD risk and to identify ways of reducing it. Moreover, they expressed the desire to address these issues with their GP or nurse. In contrast, GPs did not always perceive behaviour modification support as part of their role. Furthermore, practice nurses expressed very limited confidence to carry out these activities with patients. This means that patients are not benefitting from their attendance at CVD risk screening, a time at which the very data required to formulate a tailored risk reduction strategy is being collected and recorded by key healthcare professionals.

Discussion
This mixed-methods study of CVD risk factor assessment in people with psoriasis in UK primary care highlights important gaps in current practices around CVD risk screening. In particular, it demonstrates that CVD risk screening appears to be limited to a data collection activity rather than viewed as one component of a broader intervention strategy to reduce CVD risk. There was little evidence that opportunities for effective risk communication between patients and practitioners in consultations were recognized and acted upon by the practitioners conducting risk assessments. This study goes some way to offering a potential explanation for the debated ineffectiveness of national health check programmes and may explain Krogsbøll and colleagues' findings 5 that health checks can lead to an increased number of new diagnoses and pharmacological interventions while failing to reduce morbidity.
Our study shows that despite significant levels of risk factors identified in study participants at risk assessment (between one-third and half classified as obese, with very high waist circumference and raised BP that would warrant further investigation and almost one-fifth smoking and drinking over the recommended amounts), opportunities to support patients to understand CVD risk and/or identify risk reduction strategies may often be overlooked in consultations. A key explanatory factor was that practitioners' confidence to deliver personalized lifestyle behaviour change support was low. While this study was undertaken in the context of trying to reduce psoriasis-associated comorbidities, these findings have broader relevance to our understanding of the role of CVD screening and health checks.
A focus on the processes involved in CVD risk assessment at the level of the individual consultation provides crucial detail which aids our understanding of the apparently limited effectiveness of general health checks which have been identified at the population level 14 as well as indicating direction for more effective preventive approaches. 56 Missed opportunities in disease prevention have been previously identified in physicians' practice, 57,58 but our study supports earlier literature which states that opportunities to address change are frequently presented in consultations. 59,60 Patients participating in the current study, as in other studies, 61 wanted to discuss lifestyle issues with healthcare practitioners and furthermore expected it, 62 and addressing unhealthy behaviours is recognized as a key element in the roles of health practitioners (e.g. healthy lives, healthy people; 63 Making Every Contact Count initiative http://www. makingeverycontactcount.co.uk/; NICE public health behaviour change guidance). 64 However, practitioners express low confidence in recognizing, acknowledging and intervening with patients to manage psoriasis and its comorbidities, [65][66][67] and patients report inadequate understanding of psoriasis and perceptions of low control over the condition. 68 Known factors which assist patient perception and understanding of risk could have been used by practitioners in this study to take advantage of opportunities presented in consultations such

Strengths and limitations
A particular strength of this study was the use of T-A-R methods to go beyond self-report and enable exploration of what practitioners do in practice. This addresses a limitation identified in previous studies of CVD risk communication in primary care. 73 The study also has several limitations. Uptake of risk assessment was relatively low (possibly patients over 40 years of age had already been offered, and/or participated in a recent health check by their practice), and our sample may be unrepresentative of the wider population of people with psoriasis. The relatively small sample size means the risk factor proportions recorded may be imprecise in comparison with the general population with psoriasis. The audio-recorded consultations may not reflect routine CVD risk assessment in practice; however, practitioners were asked to carry out the process of risk assessment according to their routine practice for CVD screening and they reported very little difference between the two. Additionally, it may be that only the most confident practitioners agreed to audio recording of their consultations, however, given the low level of skilled practice observed, this may be even lower in the 'real world'. The small number of follow-ups limits findings as practitioners may have taken opportunities to address issues of risk and lifestyle in consultations that were not captured by recordings. However, the audio recordings demonstrate that opportunities for intervention present themselves at the risk assessment consultation itself, and the approaches of two practitioners in particular show that such opportunities can be capitalized upon. Lastly, as practitioners and patients were in a therapeutic relationship, social desirability may have inhibited interview accounts.

Practice implications
A shift in the focus of screening consultations is needed to encompass effective discussions and interventions to address CVD risk factors including behavioural ones. This means going beyond information/advice giving to change people's beliefs and increase motivation to make lifestyle changes and improve cardiovascular health. 43,44,74 Practitioners could take advantage of the 'teachable moment', defined as opportunities presented for them to link people's health behaviours to current health status and estimated to occur naturally in 10% of doctorpatient consultations, 75 capitalizing on these 'cueing events' 76 to prompt discussions about lifestyle change. 77

Conclusion
Screening for CVD risks is an activity which provides opportunities to engage patients in discussions about their current and future health status and offers practitioners the chance to provide timely brief interventions to improve future health. The benefits of health checks may not have materialized because of a focus upon data recording rather than intervention. Before abandoning health check programmes, it is important to attend to key process issues in risk assessment and encourage professionals to focus on helping patients understand the personal relevance of lifestyle behaviour choices, engage with the possibility of making changes and discuss individually appropriate strategies for change within the primary care consultation. and to acknowledge support for recruitment from the National Institute for Health Research (NIHR) through the Comprehensive Clinical Network.

Conflict of interests
No conflict of interests has been declared.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table S1. Measurements recorded at risk assessment. Table S2. Consultation audio recordings: framework guiding critical listening/coding. Table S3. Characteristics of patients and practitioners interviewed.