Patient, clinician and manager experience of the accelerated implementation of virtual consultations following COVID‐19: A qualitative study of preferences in a tertiary orthopaedic rehabilitation setting

Abstract Aim To investigate the experiences of patients, clinicians and managers during the accelerated implementation of virtual consultations (VCs) due to COVID‐19. To understand how patient preferences are constructed and organized. Methods Semi‐structured interviews with patients, clinicians and managerial staff at a single specialist orthopaedic centre in the United Kingdom. The interview schedule and coding frame were based on Normalisation Process Theory. Interviews were conducted over the telephone or by video call. Abductive analysis of interview transcripts extended knowledge from previous research to identify, characterize and explain how patient preferences for VC were formed and arranged. Results Fifty‐five participants were included (20 patients, 20 clinicians, 15 managers). Key mechanisms that contribute to the formation of patient preferences were identified. These were: (a) context for the consultation (normative expectations, relational expectations, congruence and potential); (b) the available alternatives and the implementation process (coherence, cognitive participation, collective action and reflexive monitoring). Patient preferences are mediated by the clinician and organisational preferences through the influence of the consultation context, available alternatives and the implementation process. Conclusions This study reports the cumulative analysis of five empirical studies investigating patient preferences for VC before and during the COVID‐19 pandemic as VC transitioned from an experimental clinic to a compulsory form of service delivery. This study has identified mechanisms that explain how preferences for VC come about and how these relate to organisational and clinician preferences. Since clinical pathways are shaped by interactions between patient, clinicians and organisational preferences, future service design must strike a balance between patient preferences and the preferences of clinicians and organisations. Patient and Public Contribution The CONNECT Project Patient and Public Involvement (PPI) group provided guidance on the conduct and design of the research. This took place with remote meetings between the lead researcher and the chair of the PPI group during March and April 2020. Patient information documentation and the interview schedule were developed with the PPI group to ensure that these were accessible.


| INTRODUCTION
Virtual consultations (VCs), a collective term for phone and video consultations, received significant interest during the COVID-19 pandemic. Their use allowed patients to access healthcare while avoiding close social contact. The COVID-19 pandemic accelerated the implementation of the NHS Long Term Plan, 1 which called for digitally enabled outpatient care across the NHS. The NHS What Good Looks Like framework 2 provides guidance for health and care leaders to digitize services with a view to 'improve the outcomes, experience and safety of our citizens.
In March 2020, the British government asked people to 'stay at home' and 'protect the NHS' as the COVID-19 pandemic took hold.
Many hospitals within the United Kingdom rapidly adopted VC to continue delivering healthcare while also adhering to social distancing guidelines. In May 2020, 185 NHS organisations were set up with the platform 'Attend Anywhere', and thousands of video consultations were carried out each day. 3 VC is now central to the ongoing functions of patient care within the NHS in the United Kingdom. VCs have been shown to result in high levels of satisfaction 4,5 and to be a feasible method to maintain care during the pandemic. 6,7 The UK Government established guidance for face-to-face (F2F) assessments during COVID-19, 8 which included requirements for risk assessments, temperature checks, face coverings, hand sanitizer, social distancing, provision of personal protective equipment, cleaning after appointments and ventilation.
The use of remote consultations before any in-person contact was recommended during the pandemic. 9 During 'lockdown', the opportunity for patients to have F2F care was limited.
Before the COVID-19 pandemic, there was an accumulating evidence base around small, pilot-stage projects of both telephone and video consultations across healthcare. A review of the literature, published in 2014, identified 27 published studies on the use of Skype (a software for video consultations) consultations with the majority of these being small pilot projects. 10 Our previously published qualitative systematic review identified nine studies reporting the use of VC (both phone and video) in an orthopaedic rehabilitation setting before the pandemic. The majority of these were small projectsembedded within larger trials. 11 The VOCAL study 12 aimed to provide an in-depth study of the advantages and limitations of video consultations across two contrasting clinical settings. Greenhalgh et al. 13 provided a comprehensive overview of the complex challenges of embedding video consultations in practice. Much of the research published since the COVID-19 pandemic investigates the acceptability of VC and the degree to which patients are satisfied with its use. 4,5 This paper is the final phase of the CONNECT Project 14 ; a mixedmethods study that investigates patient preferences for VCs. The overall purpose of the project was to understand the potential interactions between patient preferences and the use of VC in orthopaedic rehabilitation (a summary of the different components of the project is given in Figure 1). Previous phases found that patient preferences for VC are influenced by the work patients themselves are required to do, 11 their own situation and how this shapes their expectations about the use of VC. 15 Patient preferences are influenced by whether they have access to the required resources to meet the requirements of the consultation. 16 COVID-19 appeared to influence preferences in favour of a VC but we cannot be sure whether this shift is permanent. 17 This paper brings together these previous studies to develop a model of preference formation through an empirical investigation into the experiences of VC implementation due to COVID-19.
To enable healthcare services to design pathways that enhance the uptake of the appropriate use of VC in clinical practice, it is important to understand how patients form their preferences. The aims of the study reported in this paper were to investigate the experiences of patients, clinicians and managers during the accelerated implementation of VC (both phone and video consultations) due to COVID-19. The study aims to identify, characterize and explain how patient preferences to implement VC are decided and how they are organized following on from the COVID-19 pandemic. The research question for this study was 'how are patient preferences for VC decided and organised following COVID-19?' The protocol for the CONNECT Project was previously published. 14 The study is informed by two theoretical perspectives.
1. Normalization Process Theory 18 (NPT) provides an underpinning line of enquiry into the implementation process of VC. 19 provides an understanding of how patient preferences are decided for VC.

Preference theory
Both NPT and Preference Theory rely on ideas about social and mental mechanisms to explain the outcomes of implementation processes and the production of preferences. Indeed, qualitative analysis of this problem must provide accounts of why phenomena occur 20 and how these are motivated or shaped by different mechanisms. A mechanism can be defined as a process that 'brings about or prevents change in a concrete system', 21 and that involves 'constellation of activities and entities that are linked to one another in such a way that they regularly bring about a type of outcome'. 22 These definitions underpin the work that follows.

| METHODS
This paper is part of a larger body of work and forms Phase 4 of the CONNECT Project.

| Setting
The research was conducted within a single specialist orthopaedic hospital in North London, UK. All participants were recruited from within the specialist hospital. The hospital had set a target of 80% VCs 7 to reduce footfall and thus the risk of infection during the pandemic.

| Participants
We aimed to recruit 20 patients, 20 clinicians and 15 managerial staff (including operational, improvement, administrative and clinical managers). We took a pragmatic approach to recruit an accessible sample of participants: For patients, we aimed to recruit at least 10 male patients and 10 female patients; for healthcare professionals, we aimed to recruit a range of occupational therapists and physiotherapists with experience of delivering VC; for managerial staff, we aimed to recruit a range of professionals with experience of being involved with the planning, set up and delivery of VC since the start of the pandemic. Participant inclusion and exclusion criteria are detailed in Table 1.

| Recruitment
An emailed invitation to participate in the study was sent to all occupational therapists and physiotherapists with experience of using VC. Individuals within the organisation who had a role in the deployment of VC were invited to participate. Clinicians were asked to identify patients who were interested in participating. Once a patient had indicated they were happy to be approached, an email letter of invitation was sent to them, and they were asked to formally agree to be sent information about the study. Eligible and interested potential participants were provided with a participant information sheet and given at least 24 h to discuss the study with the researcher. They were enroled in the study upon informed consent, received by email, using a specifically designed email consent form.

| Data collection
The interview schedule was developed based on NPT. [23][24][25][26] Definitions of the constructs of NPT can be seen in Tables 2 and 3. The full interview schedule can be seen in Appendix S1. Interviews were conducted using phone or video call. Interviews lasted around 60 min with the option to extend or shorten as required. All interviews were

| Data management and analysis
Following transcription, the audio recordings were reviewed with the completed transcripts by AWG to enhance the familiarity with the content. The process was undertaken to review the content of the transcripts and to ensure all identifiable data were removed.
Interview transcripts were reviewed and uploaded into NVIVO (version 12). Data analysis followed the principles of abduction as set out by Tavory and Timmermans, 27 I, 11 Phase II 15 and Phase III. 16,17 The purpose of the characterisation was to abductively extend insights from the previous research to develop new insights into the development and organisation of patient preferences.
2: Codes were then characterized in relation to the research question 'how are patient preferences for VC decided' 3: Codes were subsequently characterized in relation to the research question 'how are patient preferences for VC organised' Reporting was conducted using the Standards for Reporting Qualitative Research 28 (the report can be seen in Appendix S3).  19-70 min). All interviews were conducted over video call except for two patient interviews, which took place over the phone.

| Patient and public contribution
The study interviews took place between September and October 2020, between the UK 'Lockdowns' 1 & 2 due to COVID-19.
The patients within this study were forced to have VC due to the government restrictions and local Trust policy.
This study presents significant new data and performs an integrative analysis of this in relation to old data. The integrative analysis of previous and new insights is presented in Table 2. Interview extracts of participants' perspectives may be found in Table 3.

| Coding and integrative analysis of interview data
Interview data were coded and characterized in relation to the previously identified factors that influence preference, identified from our earlier research. New insights were identified during this process.
The integrative analysis led to the identification of factors that shape the formation of patient preferences for VC and are described below.
The knowledge underpinning these factors from our previous research and new empirical data within this study are presented in Table 2. 3.1.1 | The context for the consultation The context for the consultation is the circumstances that form the setting. This includes the expected standards and rules of care (normative expectations), the expected ways patients and clinicians are organized and relate to each other (relational expectations), the degree to which features of the consultation meet the requirements of the consultation (congruence) and the access to material and cognitive resources to support the consultation (potential).

Normative expectations
Patients' expectations were founded on their previous experience of care. All patients within this study had experienced in-person physiotherapy before and were able to speculate about the effectiveness of VC. The requirements of the consultation provided a reference point to understand the way VC would work for them. During COVID-19, 'stay at home' became law and patients were satisfied with virtual care during this time and many were happy to not travel.
The presence of COVID-19 led to VC becoming the only way to access rehabilitation for the majority of patients and during this time patients in this study preferred VC to no care at all.

Relational expectations
Patients had expectations about the ways patients and clinicians relate to each other during clinical interactions. Their previous experience of care provided a reference point to understand the changes in relationships with their clinicians over VC. Although many patients felt interactions over VC were inferior to F2F care, patients were willing to compromise and accept VC during COVID-19.

Congruence
The clinical status of the patient and the treatment required

Potential
Patients' access to resources shaped their ability to engage with virtual care. These resources included hardware (such as a phone, tablet or computer) and software (such as up-to-date operating software and the platform to undertake a video call). During the COVID-19 pandemic, the platform Attend Anywhere was made available across the NHS in England. Resources were made available to patients to support the use of video calls.

| The implementation process of VC
Participants within this study were not offered the choice of a F2F consultation and all had to implement VC (either a telephone call or a video call with their clinicians). In these circumstances, a process of implementation took place. NPT provided the framework to build on previous iterations of the CONNECT Project to explain the implementation process for patients. 18 Coherence Patients needed to understand the differences between VC and F2F.
This was challenging during the pandemic when the introduction of VC was accelerated and individuals were inexperienced in VC as the main form of consultation. Clinical and administrative staff supported patients to understand the role of VC. The capabilities of VC were seen to be limited where an in-person intervention was required, such as when hands on-manual therapy or facilitated exercises were required. If a patient was concerned about their problem, they often felt that a thorough F2F assessment was preferable to a VC.

Cognitive participation
In general, patients who found F2F attendance challenging were more committed to VC. For some, a traditional F2F appointment took significant planning and left the patient in pain due to their travel.
Commitment for VC was enhanced with increased congruence for the patient. Many patients were concerned about catching COVID-19 through travel to the hospital and this made the option of a VC preferable. Patients' willingness to use VC was shaped by their understanding of the benefits.

Collective action
VC rehabilitation was challenging in the home environment for some patients. It was not possible to conduct the range of interventions that were often needed if the patient's video device was not portable.
Mobile devices were helpful if, for instance, a patient had to film themselves walking upstairs or an occupational therapist needed to observe functional activities in the kitchen. Patients had to convey their symptoms over VC without the clinician being able to physically touch them.
The 'work' required of patients and clinicians over a VC was different from the 'work' of F2F care. Some patients and clinicians did not have the technical skills required to be able to use VC. Family members often supported patients with VC activities. Clinicians occasionally needed to teach patients the required computer skills over the phone. The burden of VC shaped preferences for ongoing use of VC.

Reflexive monitoring
Patients were forthcoming with feedback about their experiences.
Clinicians also discussed their own experiences to shape the virtual service. For instance, after several clinicians encountered technical challenges that interfered with the delivery of a VC, the virtual slots were increased from 30 min to one hour. Patients valued the extra time with their clinician and found this aspect of the VC to be beneficial. In response to these technical problems, clinicians made it clear to patients, at the start of a video call, that they would contact the patient via telephone if the VC cut out. As patients and clinician dyads experienced both VC and F2F, they were able to plan longterm management, which often included the use of both VC and F2F.  Patients' access to material and cognitive resources shaped the way in which they made sense of their responsibilities and the value of the alternatives, as well as their willingness to implement them.
There was recognition that different individuals would have different access to resources. It was this level of access that shaped patients' ability to do the work of the alternatives. Some patients had access to adequate broadband and a device to be able to undertake VC and some had access to equipment and the space to be able to complete their rehabilitation in the home environment. Without these, successful implementation of VC was not possible and patients were more likely to prefer a F2F.

| How preferences for VC are organized
Patient preferences were formed in the context of clinician and organisational preferences. The clinicians within this study were required to implement VC at a pace that required restructuring of policies and procedures. For many clinicians, the addition of VC worked well whereas for others VC was inferior to F2F.
The organisation invested heavily in resources for clinical staff to be able to undertake VC with patients. These additional resources sources were fixed and available alternatives for patients were restricted. When a clinician did not think that a VC would work, they would suggest a F2F, which influenced the patient's sense-making of the alternatives and their commitment to VC. The work of implementing the alternatives for patients was shaped by the resources they could bring to bear; if they did not have access to adequate equipment, they were unable to do the required work to implement VC. In some circumstances, clinicians did not believe VC was appropriate.
Organisation and clinician context, the availability of alternatives and the work required of implementation directly influenced patient preferences and decision making.

| DISCUSSION
This qualitative interview study is underpinned by NPT 18 and Preference Theory. 19 This study has extended the findings of our previous research through an investigation into patient, clinician and manager experience of the accelerated implementation of VC.
day. So, I think slowly we'll build that up. I think yeah, we've been more cautious.
[C4 -16] I thought they were in this place and I thought they were doing this and exercise z and I saw them and they were worse than I thought they were. That has also frightened peopletherapists I guess, thinking that, oh I thought they were better. One of those patients is coming back to see us as an outpatient, as a face-to-face. I don't think that it did meet her needs, actually, from a pain-I think she needed to be taken out of her environment which is quite challenging, quite chaotic and quite toxic at times.

| Strengths and limitations
A strength of this study is the cumulative abductive identification of insights through the different phases of the CONNECT Project, before, 11

| Mechanistic model of preference formation
Here, we present a theory of preference formation. A visual model to illustrate the formation of preferences has been developed from the integrative analysis and can be seen in Figure 2. We consider the formation of patient preferences as a mechanism. Our position is that patient preferences are the product of a total subjective comparative evaluation of the available options. The context for the consultation (normative expectations, relational expectations, congruence and potential), the available alternatives and the implementation process (coherence, cognitive participation, collective action, and reflexive monitoring) are all involved in shaping the total subjective comparative evaluation. These are the key entities that are linked to one another to form the construction of patient preferences.

Consultation Context
Each individual patient context will present a unique potential to incorporate either a VC or a F2F for a clinical appointment. For some patients, the use of a VC will be burdensome; for others, the introduction of VC will be beneficial. Patients will need to have access to specific resources (the required hardware, software and skills to use these) 30 to have a VC, particularly if VC is enforced.
Patients will also need to be prepared to accept the change in

The formation of preference
A total subjective comparative evaluation is undertaken by the patient. The patient will consider all the available information and choose the alternative, which brings them the most benefit. The patient will prefer the option that yields the most benefit.

The consequences of choice
The choice a patient makes will have a range of consequences on their context, their implementation process, and their overall preferences. The outcomes and consequences will differ for each individual patient, as this is all dependent on their individual context.
A patient is more likely to implement a preferable alternative of care. This understanding of the mechanisms that influence preference formation is helpful to understand implementation processes. Patient satisfaction is positively associated with technical performance, 33 and in our study, clinicians often had to support patients with technical challenges. Some patients did not possess the technical skills to use VC, 7 which reflects the nationwide picture. 30 While this clinician support may have a positive impact on patient experience, this will reduce the overall resources of the clinical team to be able to provide rehabilitation for patients.

| Results in context
Communicating over VC placed greater emphasis on verbal communication skills during these interactions. 34 Failed VC was deemed to occur when there were issues with communication. 6 In addition, clinicians needed to be able to trust the VC-many orthopaedic professionals lost confidence with virtual calls when issues arose. 35 Clinicians' normative expectations of undertaking a thorough hands-on assessment were important, many feared missing sinister pathology and screening of 'red flags' 36  Organisations and clinicians have a role in helping patients to understand the role of VC and some of the ways in which organisations and clinicians can influence patient preferences are shown in Table 4. The application of preferences and decision-making may take place as a shared decision, where patients and clinicians have equal power, or the more powerful individuals may exert their own preferences to enable preferable outcomes ( Figure 3). Consideration of these mechanisms will facilitate shared decision-making in practice.
The NHS Long Term Plan 1 set out a vision for a digital NHS but the COVID-19 pandemic led to a 'big bang' of technological change 43 where services rapidly converted F2F to VC in line with government guidelines. The timescale for the relaxation of social distancing restrictions in the UK remains uncertain; the capacity for F2F clinics will continue to be reduced during this period. Predicted modelling suggests that up to 28 million operations were cancelled or postponed globally during the first wave of COVID-19 44 and orthopaedics is now facing a substantial backlog of surgical cases. 45 There is likely to be an ongoing reliance and pressure to use VC as remote consultations have been proposed as a potential way to increase capacity in orthopaedics. 46 This pressure will continue to influence clinician and patient preferences. Healthcare must, therefore, be sensitive to clinician and organisational preferences. Clinicians need to develop sensitive ways to manage the 'arenas of struggle' 47 between highand low-powered individuals when preferences are incongruent.
Agreements between healthcare professional and patient preferences are more likely to lead to successful uptake and adherence to modalities that patients conclude to be more beneficial.
Within our theoretical model, a patient will prefer the alternative that brings them the most benefit.

| CONCLUSIONS
This was an empirical investigation into the experiences of patients, clinicians and healthcare managers during the accelerated implementation of VC during COVID-19. This study has explained patient preferences through the accumulation of several pieces of work as VC changed from an experimental clinic to a compulsory form of service delivery during the COVID-19 pandemic. This study presents a robust conceptual model of preference formation.
Patient preferences are decided in the form of a total subjective comparative evaluation of the available alternatives of care. This study found that the implementation process of investing meaning, commitment, effort and comprehension into the available options informed the total subjective comparative evaluation and the formation of preference. The preferences of clinicians and the organisation need to be considered as these were shown to mediate patient preferences. Since decision-making will take place in the context of patient's, clinician's and organisations' preferences, future pathway design should be sensitive to patient preferences while acknowledging the preferred outcomes of clinicians and organisations.