Anything but engaged: user involvement in the context of a national electronic health record implementation.

Background The absence of meaningful end user engagement has repeatedly been highlighted as a key factor contributing to ‘failed’ implementations of electronic health records (EHRs), but achieving this is particularly challenging in the context of national scale initiatives. In 2002, the National Health Service (NHS) embarked on a so-called ‘top-down’ national implementation strategy aimed at introducing commercial, centrally procured, EHRs into hospitals throughout England. Objective We aimed to examine approaches to, and experiences of, user engagement in the context of a large-scale EHR implementation across purpose-fully selected hospital care providers implementing early versions of nationally procured software. Methods We conducted a qualitative, case-study based, socio-technically informed, longitudinal investigation, purposefully sampling and collecting data from four hospitals. Our data comprised a total of 123 semi-structured interviews with users and managers, 15 interviews with additional stake-holders, 43 hours of non-participant observations of meetings and system use, and relevant organis-ation-speciﬁc documents from each case study site. Analysis was thematic, building on an existing model of user engagement that was originally developed in the context of studying the implementation of relatively simple technologies in commercial settings. NVivo8 software was used to facilitate coding. Results Despite an enduring commitment to the vision of shared EHRs and an appreciation of their potential beneﬁts, meaningful end user engagement was never achieved. Hospital staﬀ were not consulted in systems choice, leading to frustration; they were then further alienated by the implementation of systems that they perceived as inadequately customised. Various eﬀorts to achieve local engagement were attempted, but these were in eﬀect risk mitigation strategies. We found the role of clinical champions to be important in these engagement eﬀorts, but progress was hampered by the hierarchical structures within healthcare teams. As a result, engagement eﬀorts focused mainly on clinical staﬀ with inadequate consideration of management and administrative staﬀ. Conclusions This work has allowed us to further develop an existing model of user engagement from the commercial sector and adapt it to inform user engagement in the context of large-scale eHealth implementations. By identifying key points of possible engagement, disengagement and re-engagement, this model will we hope both help those planning similar large-scale EHR implementation eﬀorts and act as a much needed catalyst to further research in this neglected ﬁeld of enquiry.


Introduction
Information technology (IT) is increasingly being utilised to facilitate the sharing of information across teams and groups of healthcare staff. 1 In the hope of realising more effective and safer care, many countries are actively pursuing the implementation of electronic health record (EHR) systems through making major investments in these initiatives. 2However, the existing literature suggests that many IT implementations, particularly those that involve complex organisational transformations, fail to realise their full potential, this often reflecting, amongst other things, the lack of effective approaches to user engagement. 3,4o date, agreeing on a shared definition of user engagement in the context of technological innovation in healthcare settings has been difficult, as approaches and contexts vary significantly.Its conceptualisation often depends on the technology in question and the need to accommodate alternate perspectives of different stakeholders, including both implementers (often focusing on the process of engagement) and user groups (often focusing on subjective experiences of this process). 5][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] The underlying assumption here is that users are best placed to understand the intended context of system use; by contrast, developers and implementers may lack important clinical insights, resulting in a lack of understanding of the potential consequences of technologies for end users. 26,27ngland was one of the first countries to make substantial efforts to implement procured EHR systems into hospitals on a national scale.The National Health Service Care Record Service (NHS CRS) was part of a wider program to modernise the NHS through the National Programme for IT (NPfIT).The NHS CRS has been conceptualised as a 'top-down' implementation, this reflecting its central management, substantial scale and ambitious implementation timelines. 28,29NHS Connecting for Health (NHS CFH), an 'arms-length' government agency, was charged in 2005 with implementing these nationally procured EHR systems.These included iSOFT's Lorenzo Regional Care (henceforth referred to as Lorenzo), Cerner's Millennium and CSE Healthcare's RiO.
In this paper, we focus on the approaches to, and experiences of, end user engagement in relation to the implementation and adoption of Lorenzo software (which is described in Box 1).We conceptualise user engagement as a form of involvement which will lead to informed implementation of an effective system that is assimilated into working practices due to close alignment with user needs and expectations.][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] Although such co-creation models have been studied previously, the implementation as part of NPfIT was of particular interest as this differed from the smallerscale implementations studied hitherto and offered potential insights into user engagement within a multiorganisation and a multi location context.
We thus seek in this paper to build on the existing user engagement literature in relation to IT in health care by reflecting on the approaches to and experiences of user engagement in a national implementation of complex EHRs.][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] We draw in particular on an existing theoretical model of user engagement from less complex, commercial IT applications (Figure 1), 5 and our own work to explore approaches to user engagement in the context of the NPfIT.Based on our findings, we map out where the differences between national implementation and Box 1 Characteristics of Lorenzo Regional Care implemented as part of the NPfIT .Lorenzo is a specific type of web-based EHR software that is built whilst being implemented in the North, Midlands and Eastern Region (NME) of England covering 60% of the country. .It was originally planned to be implemented as a single solution across both primary and secondary care settings, but the scope was subsequently reduced to exclude primary care settings as contracts were repeatedly renegotiated in order to reduce costs in a climate of increasing economic uncertainty. .Lorenzo does not exist as yet in its full form, as the original intention was to develop a system in collaboration with the NHS so that it would address the needs of users.Different releases are available as soon as they are developed in India, where most of the developers are based. .Although releases have to be implemented consecutively, organisations are to some extent free to choose which parts of releases they wish to implement according to their needs.
single organisation-centred implementation approaches and associated user engagement lie, how engagement efforts in relation to the NPfIT have been perceived by users and organisations, and how approaches to facilitate engagement can be applied more effectively to large-scale IT implementations.We conclude by offering some preliminary recommendations arising from this work that national and international policy makers and implementation teams may wish to consider.

Design
For the purposes of this paper, we drew on a subset of qualitative data collected as part of our national evaluation of the NHS CRS in English hospitals. 28,30,31n doing so, we focused on the implementation of Lorenzo as a complex type of EHR software that was intended to be co-created in collaboration with NHS staff, and which therefore did not exist in its final form when implementation began.

Data collection
Our dataset comprised a combination of 43 hours of observational fieldwork and semi-structured interviews with 123 stakeholders including users and managers from case study sites.These were complemented by interviews with an additional 15 stakeholders outside hospitals including governmental stakeholders, developers and representatives from the independent sector.We also collected and analysed a range of associated hospital-specific (an average of three in each case study site) and national documents, which we treated as secondary data sources.Documents, interviews and observations at case study sites allowed us to investigate how the national implementation was approached by local management and received by local users.Examination of national documents and interviews with a wider range of stakeholders gave insights into the broader, national landscape in which these developments were taking place.
Where possible, data from four hospitals were collected at two different time points, approximately six months apart, to allow a certain amount of maturation of software development.During this time, organisations had also expanded their user base and/ or software functionality.This longitudinal data collection allowed us to capture developments over time as organisations and users worked out the consequences of the nationally implemented system with incrementally increasing functionality.

Data analysis
Data were collected and analysed by a designated lead researcher who had overall responsibility for data collection at all four case study sites (KC), drawing on the approach outlined by Miles and Huberman, 43 and Mason. 44Interviews were transcribed and together with documents and field notes uploaded into NVivo8 software.Initially, data were organised along dimensions identified in the literature as important for 'successful' EHR implementation, whilst still allowing new categories to emerge.Initial coding dimensions included technical, human/social, organisational and macro-environmental factors. 45Across these dimensions, we examined issues relating to user engagement in more detail by retrieving data from all sources coded against user engagement.We then examined the data in this category and developed subthemes, initially within and then across case study sites and at different time points. 43These were refined based on seeking complementary contextual data (providing wider contextual insights into the situation), confirmatory data (supporting prior theoretical assumptions developed from other sources) and disconfirming data (those that did not fit with developed theoretical explanations).Disconfirming data and inconsistencies between data sources were examined in most detail.The use of matrices facilitated this analytical process.Throughout, the approach was to investigate user experiences and management efforts through an interpretative lens, 44 seeking to understand how the new system and associated engagement efforts were perceived on the ground.

Results
A full summary of our dataset and a brief description of each case study site are given in Table 1.][47][48] We have summarised these graphically, building on an existing model of user engagement from less complex commercial technologies in Figure 2. 5 We identified the following subthemes, which will be considered in turn: . layers of complexities to engagement approaches resulting from the national procurement .usability and customisability issues .the role of champions and other key individuals .the complexity surrounding the hierarchical structures and associated engagement efforts of clinical staff.
We begin by describing the strategies for user engagement employed within the context of the national implementation as a whole and in our case study (Lorenzo) sites in particular.We then describe experiences of user engagement efforts on the ground, before discussing potential ways forward.

Understanding approaches to user engagement in the context of a national implementation
Users generally bought into the overarching vision of nationally shared EHRs, thereby providing a receptive basis for the initial 'point of engagement' (see Figure 2).
'Electronic I think it needs to be done now, I think.I don't know, I just think the day of paper notes is probably gone when there's so much technology around...If you think about it's a very ancient way of doing things to write everything down when there's so much technology out there ... Good vision, but whether this system could do it I don't know.' (Interview, healthcare professional) Naturally, this overall vision encompassed a number of expectations based on existing needs and the hope for the new system to address these: 'The expectation of the service that I had, I mean I went to a launch a couple of years ago and when they launched it.It was like ''wow how cool would that be if you could put in a number and the whole history of someone came up'', especially because from the [name of area] we have a lot of people from away as well and you can actually see what, all that data and all that information so you get to know the patient quicker because sometimes the paper notes take User involvement in a national EHR implementation 195 time to come through and things can be done accordingly.'(Interview, healthcare professional) However, the vision of shared EHRs became compromised by national arrangements.Although the government was a significant stakeholder in driving the overall implementation, it did not directly facilitate the engagement of users as this responsibility was largely devolved to individual organisations implementing Lorenzo.As hospital staff were potential users, hospitals were considered to be in a good position to coordinate local engagement efforts.As one confidential national document reads: '[Engaging clinicians] is seen as a key issue, and any difficulty experienced in engaging with clinicians will result in a reduction in the speed of the implementation and in benefits realisation ... [the] devolved approach to implementation is rooted in the belief that one can only implement changes to working practices at a local level.The focus for clinical engagement is therefore at that local level....' Local engagement efforts were, however, complicated by the 'top-down' implementation and the nationally procured nature of Lorenzo software.Here, a close working relationship between users and management introducing the change was not possible as implementation was led by national structures.Local organisations (including users) were not involved in systems choice, which should have been an important consideration at the initial 'point of engagement' (i.e. the beginning of the process of engagement), building on the common vision of a national EHR (Figure 2).This is illustrated by the following extract from a Department of Health publication: 'The Department [of Health] ... decided to conclude the bulk of procurement activities before focusing on communicating with and engaging NHS staff.Wider engagement and mobilisation of the NHS was not started until [it was] judged that procurement had reached a sufficient stage of maturity to be able to communicate its outcome in a meaningful and efficient way.It was concerned that to have done so earlier might have raised expectations which were either speculative or may not have been met and there were also resourcing constraints.'(Source: Department of Health, 2006) 49 As a result, hospital management found itself trying to 'sell' software that it had not chosen to their clinical and administrative staff.This was a difficult undertaking as early release software had limited functionality and offered little in the way of benefits to clinicians or patients, and was replacing well-functioning local systems.In addition, they could not demonstrate the product to its potential users as it did not exist in its final form at that time.These difficulties are exemplified in this interview extract: 'I'll never forget this, when we had a Programme Board ... and they said we have been told by the Secretary of State Over time, hospital management therefore tended to lose credibility amongst users.Engagement was in effect inhibited right from the start as the new system could not be demonstrated to users, with the consequence that changes in business, clinical and administrative processes could not be planned for.The first quote below illustrates this lack of credibility, whilst the second quote illustrates how problems in business planning were perceived to impact on efforts to engage system users: '... one of the difficulties for us as a Trust was that we had to design our business processes without having access to the system and I'm sure that's something that's been said throughout all the other Trusts is, you know, when you're engaging with clinicians because it's, you know, it's being sold as a clinical product and you can't show them how it's going to work it's very hard for them to say well, yea, we'll use this bit here and we'll use that bit here and so for me it was a challenge.'(Interview, manager) Problems with the lack of credibility of the software were further exacerbated by early negative experiences with Lorenzo, with clinicians expressing concerns around increased workload, and hospitals reporting that it negatively impacted on organisational functioning.Consequently, users began to disengage with the system implementation (see Figure 2).
'It does take a little longer to request an investigation than it does using a pen and paper and clearly if we're going to Figure 2 A model of user engagement emerging from our results use an electronic system it needs to be at least as efficient in terms of time utilisation as the pen and paper otherwise it isn't going to get used universally.So that needs to be improved a little bit but I think it is at an early stage and there's a lot of potential to the system.' (Interview, manager) Relevant national bodies (in particular NHS CFH) and local management (hospitals) attempted to address this lack of user engagement with 'engagement strategies' focusing on 'stakeholder management' predominantly aimed at clinical users.These approaches were designed to mitigate a recognised risk of users refusing to use the new system.They included targeted communication of anticipated benefits of usage, training, the appointment of clinical leads and attempts to increase users' input into system design. 9,14,50,51Hospital management's targeted communication strategies were illustrated during the following interview: 'I think it's about what are we communicating, how are we communicate that, what are we saying our expectations are to staff about what are the benefits of this system to patients and to staff.How can we sell that and we need to start developing our, we've got a communication plan but we need to start developing that now, we need to start rolling it out.So we've got, clinicians, doctors who want to use something that's going to add to their working day not something that's going to add more time and more output.'(Interview, manager) Such local engagement strategies within hospitals were intended to provide a 'push in the right direction' through attempts to win over sceptical staff by emphasising the likely benefits of use.This sometimes also involved a certain amount of spin, as the following extract indicates: '... it's all about making them [referring to users] feel valued because ... we've got this.It might not always be entirely the truth, you know, but it makes people feel oh well yeah and it just makes people that little bit more compliant to try it, it's a bit of bribery, it's people management.' (Interview, manager) These 'soft' strategies appeared to work in some instances, but if they were not successful, the next management step was to mandate use of the system.The approaches employed to enforce systems usage varied but included withdrawing existing paper systems.
'I mean there's the carrot, the stick and then a cattle prod and, you know, I mean sometimes you have to use a combination of all three, you know, you wheel and cajole, you make the technical solutions as easy as possible but, you know, people generally don't want to change and until you force them to change by taking away their paper they won't change.'(Interview, manager) However, in keeping with Figure 2, these strategies targeted stages past the 'point of engagement', i.e. once the implementation was already well underway and users were obliged to use a system that had been chosen for them by 'the top'.Most efforts were therefore focused on targeting disengagement and re-engagement stages (Figure 2).

User experiences and perspectives of engagement initiatives
Our results illustrated that user experiences of this technology implementation and associated engagement initiatives had significant consequences for user engagement.Despite widespread agreement on the vision, many users became progressively more disengaged over time (Figure 2).This was most likely shaped by a lack of system usability and customisability; a lack of effective clinical champions and other key individuals that could span boundaries between users and management; and the complexity of the work environment, its hierarchical structures and associated engagement efforts aimed particularly at clinical staff.
In relation to Figure 2, strategies to address issues with disengagement can be placed at the engagement and re-engagement stages representing mitigating actions by management.We will discuss each of the issues and associated user experiences in turn.

Usability and customisability
Lorenzo was designed as it was being implemented, which potentially allowed significant user involvement.However, system choice and local customisability were limited for reasons of large-scale interoperability, constraining changes that individual users and organisations could make to the system.
'I think all the correct elements are there but, you know, I don't know really how flexible the system is but I know there has been some medical input into making it user friendly for clinical teams, but I think, my feeling is it would benefit from more input from clinicians so that we can get an output that is useful to us.And at the moment we're not quite there really and particularly with things like the generated discharge summary, it's quite a lengthy document, it comes out at four sides of A4.' (Interview, healthcare professional) Here, national arrangements and associated layers of bureaucracy, including a range of governmental and commercial committees, meant that technical issues reported by users often remained unresolved for extended periods.This led to users feeling that they had not been listened to and contributed to disengagement.'... you never get, you never get consulted on anything it's just you're doing it and I think that's what gets people's User involvement in a national EHR implementation 199 backs up really, it's just that you're just expected to do it and there's no negotiation or, you know, this is why we're doing it or anything, it's just it's here, get on with it.'(Interview, healthcare professional) This was compounded by the difficulties experienced with integrating Lorenzo with existing care practices.During the period of our data collection, most users reported that it caused them additional work without bringing the promised benefits.
'I think it would be a great system and I just don't think it works very well.I don't know if it's an appropriate system.It seems to have a lot of downfalls.But I can actually see the bigger picture that yes, it would be really good.I just think we are struggling a little bit with it.'(Interview, healthcare professional) As a result, users often did not actively participate in communication, engagement and re-engagement efforts (which we refer to here as strategies that are designed to address recognised issues with disengagement) initiated by management, a situation which may have been exacerbated by concurrent changes in the health service and associated 'change fatigue'. 52ith all the other changes that are taking place in the health service I think it was just another thing that just people think if you ignore it, it will go away and of course that's not going to happen but that's quite often a perception in the health service with it being so big and there's much red tape that you can avoid things.'(Interview, manager) Missed opportunities at the 'point of engagement' (Figure 2), together with a lack of systems functionality and performance, undermined subsequent engagement and re-engagement efforts.A failure to demonstrate to users that their concerns were being listened to and acted upon impacted on users' willingness to invest time and effort in making the new system work and on the credibility of engagement efforts, which were described by some users as a 'fac ¸ade' and 'tokenistic', designed to persuade them to use a system that was viewed as lacking fitness for purpose.This contributed to disengagement, resistance and a feeling that there was no real involvement in decision making.
'... but we're only little cogs in a little wheel so they won't listen to us so ... yea, we don't get listened to ... I'm still waiting three weeks down the line to get my fax machine, to get a gateway on my fax machine so I can start faxing again properly cause our fax machine they gave us in the first place wasn't fit for purpose cause it only does thin paper and the cards we have to fax are thick, I mean just something as simple as that....' (Interview, healthcare professional)

Champions and other key individuals
There is evidence that the appointment of clinical leads and 'boundary spanners' can be effective as these individuals often have an insight into 'both worlds': management and clinical. 10,42,53Indeed, we found the use of such local champions to be valued by system users and hospital management.As a result, their expertise and influence were harnessed where possible.
'And so you don't go in and say right we're going to do business process management, you know, you have to facilitate it in such a way that you're using their language, that you can convert, you can translate.So I was almost like an interpreter for them in terms of, you know, no you don't go in and say that, don't you dare do that to them, leave them alone I'll do this bit.And there's also something about the clinicians, either they rate you or they don't, you've either got the credibility or you haven't and I think that was quite important.'(Interview, healthcare professional) However, centrally appointed, national champions were viewed by some users as lacking credibility because effective two-way communication between those individuals and users did not occur.Similarly, a number of users stated that some clinical leads, despite their clinical background, did not seem to be connected closely enough to those that they were appointed to represent (i.e.clinicians).
'I mean they talk about having clinicians as part of the developers but they're clinicians that haven't been clinicians for such a long time.I mean there's [Name] what's his name who's ... Who's an anaesthetist or was an anaesthetist but when was the last time he ever had to input anything on a computer to do anything with a patient, probably never.' (Interview, healthcare professional)

Clinical engagement versus user engagement
We further found that, despite a strategy of engaging clinical staff, guided by local management assumptions that if consultants could be 'won over' then other staff groups would follow, senior consultants did often not engage in local implementation discussionspossibly because they had already reached the disengagement stage (Figure 2).'Well yeah consultants are kind of, they don't come to events you go to them and it's making sure you go to the right events and get the right sort of message or you get one or two consultants, they don't need to be enthused about the benefit ... they just need to understand the agenda and have a view, it might be a negative view but at least they're talking about Lorenzo.So we're actually starting to engage with what we would class as the senior stakeholders across the two sites, deliberately engaging consultants...' (Interview, manager)  This may be partly due to the nature of consultants' work environment, where the timely delivery of patient care often took priority.
'Clinicians are also, if they change their working practice to use these new systems then their tolerance of failure would I think be a lot less than maybe an accountant or a traditional user just because of the nature of what they're doing.(...) I don't mean tolerance as in getting cross, they need access because the sort of environment they're in, then if something isn't working and their process depends on it, you can't say to the patient, ''Just sit there for 20 minutes while I hang on at the service desk, everything's going to be fine''.'(Interview, manager) The important role of senior consultants in the hierarchical structures within health care to some extent justifies the focus on clinical engagement, as does the fact that they can be both users and managers.This approach to engagement was therefore based on the idea of opinion leadership, with consultants mandating local use.However, consultants delegated many responsibilities, including data input into Lorenzo, to their juniors.These arrangements created a situation in which consultants were often the clinicians with the least exposure to using Lorenzo.
'I think it's more to do with the hierarchy of the clinical team in that the more junior you are in that clinical team the more of the admin stuff that you get to do.Or you're asked to do, you're expected to do.And a lot of that admin stuff is documenting in the clinical notes.So acting as scribe on the ward round or being asked to place a request for a patient or whatever and they're the duties that Lorenzo supports.So the junior doctors have got more exposure to the system earlier than the senior members of the staff which then creates its own problem in that the seniors are then more reluctant to expose themselves with a new product...' (Interview, manager) In addition, our results indicated that the engagement of non-clinical stakeholders, who were often the most frequent users of the early Lorenzo functionality, received far less attention.As a result, these users were often disengaged, disillusioned and frustrated.

Summary of main findings
This work has enabled us to describe and understand the consequences of the various engagement approaches employed in the context of the national implementation of Lorenzo.The longitudinal nature of this work has furthermore allowed us to appreciate how this evolved over time, which has been important in facilitating the appropriate adaptation of the existing, commercially orientated model of user engagement for use in the context of studying large-scale EHR implementation (Figures 1 and 2). 5 Despite the presence of an overall vision at the 'point of engagement', local engagement efforts have been negatively influenced by a lack of user involvement in procurement decisions as well as implementation timelines, and the nature of the Lorenzo system itself, which was perceived to lack usability and customisability.These factors resulted in notable disengagement of users.The re-engagement efforts we observed were often approached as risk mitigation strategies to prevent further alienation of users, but such strategies had already missed an important opportunity to initiate engagement (i.e. the 'point of engagement').Drawing on local clinical champions and boundary spanners was found to have the potential to facilitate re-engagement to some extent, but non-clinical staff (often the most frequent users of the functionality) were particularly neglected in relation to engagement and re-engagement efforts.

Strengths and limitations of this work
Our results have built on the existing literature, which to date has not considered the complex issue of user engagement in the context of national EHR implementations (see Figure 2).By researching sites longitudinally, we have been able to describe local and national engagement strategies employed and understand the effects of these on users, and the potential mechanisms involved.
However, our work is not without its limitations.We have, despite being able to trace developments over time, investigated the early stages of Lorenzo implementation only, resulting in limited insights into the more embedded use of the system.In addition, our clear rationale for focusing on one type of national EHR system due to its unique features (i.e.Lorenzo) means that the transferability of our findings to other EHR systems and contexts would benefit from further consideration.

Considering our findings in relation to the wider literature
Our results have shown how a national 'top-down' EHR implementation conflicted with the notion of user engagement itself, illustrated by political pressure to implement centrally procured systems, whereby users lacked system choice and customisability.In line with this, other authors have argued that participation in government initiatives may not be as participatory as it may first appear; eventually certain governmental objectives need to be achieved, with the result that these objectives may at some point become 'too important' to be participatory. 54][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] Such efforts should focus on the 'point of engagement' building a solid basis for the future, as opposed to targeting the disengagement and re-engagement stages.That said, we accept that any approach to engagement needs to be tailored to the local situation, balancing the requests of users with organisational and strategic requirements.For example, it is likely that some user requests will contradict others, or requests may adversely affect organisational functioning.This therefore necessitates a careful balancing act of diverse priorities by the implementers.
The literature suggests that user engagement is most effective if the system is 'home-grown' and customisable. 45,55More typically in such scenarios, a local need is identified and users themselves slowly, often over a period of decades and supported by local management, change the system to address this need. 56ere, engagement is inherent in the development process as implementation is driven by users.However, a small-scale, evolutionary approach does not address large-scale interoperability.It is also becoming less affordable and in many cases felt to be less attractive because of the slow pace of development. 56verall, effective user engagement in large-scale EHR implementations is complex and in some respects even contradictory.The question of how users can be involved whilst at the same time achieving some degree of system interoperability remains, but our work has highlighted the importance of the initial 'point of engagement' and the effect of 're-engagement' strategies.In an ideal scenario, the word 'engagement' would hardly be mentioned as it would simply 'be there' without the need for any mitigation strategies to re-engage disengaged users.Our study supports previous research which found that the issue of engage-ment only seemed to become important when it was identified as a risk to implementation 'success', or offered a retrospective explanation for an implementation being perceived to be a 'failure'. 57,58Users may be primed to this and labels such as 'stakeholder engagement strategies' may therefore arouse such questions as: 'is there any reason why I should not be engaged?'.This could further undermine the concept of engagement and contribute to user alienation and resistance to implementation.

Implications for policy, practice and research
It is not within the remit of this paper to debate whether a national EHR implementation approach is an optimal strategy, so we concentrate our thoughts on how to facilitate engagement from this point forward (see Box 2 for a summary).Our suggestions may also be transferable to other large-scale IT implementations in the healthcare sector.
In line with the new strategic IT direction of the UK government, with a growing emphasis on local systems choice, 59 we expect some of the problematic complexities of a centralised, national approach to diminish in future.For example, devolving of engagement responsibilities to individual hospitals is likely to continue and to be facilitated by more local input in systems choice and tailoring of local systems.However, it is important to recognise that future implementation efforts will be likely to require continuous re-engagement strategies as the 'point of engagement', initially promising as it built on a common vision, has received far too little attention.The English implementation context will probably therefore continue to be characterised by retrospective risk mitigation strategies in relation to engagement and re-engagement, as opposed to the preferable focus on the initial 'point of engagement'.

Box 2 Recommendations emerging from our work to facilitate user engagement in largescale IT initiatives in health care
The implications for similar ventures internationally may also be considered.First, a focus on the initial 'point of engagement' is vital.This can be achieved by building on a common vision and user input in systems choice, which may have to be local by definition.Second, there is a need to realise that if the system being implemented is perceived as inadequate and has limited capability for customisation to local needs, engagement strategies are likely to fail.Management focus should therefore be not on engagement per se, but on finding a solution that works for as many members of the organisation as possible to improve business processes as well as ways of working.Conversely, trying to impose a new computer system that is perceived as inadequate by users without the opportunity for it to be changed in appropriate timescales is likely to act as an insurmountable barrier to achieving any degree of user buy-in.
In addition, we argue for the need for user representation to go beyond a on medical consultants to understand and promote the views of diverse staff groups involved in the delivery of health care.The participation of representatives of clinicians, administrative staff and patients, in numbers that reflect the scale of the planned eHealth implementation, should enhance meaningful user engagement and reduce the challenges to implementing.Furthermore, this could enrich system functionality through closer alignment with the needs of the service for which it is intended.This form of engagement undoubtedly takes time: users' time, the time to communicate with and listen to users, time to revise implementation strategies if necessary and time to customise the product.In addition, sufficient resources will be required to facilitate the availability of users to participate in engagement initiatives whilst fulfilling the requirements of their employment.
We also accept that engagement strategies may lose momentum for reasons that may not be anticipated.In these instances, there are strategies that can help reengage users.For example, having a dedicated individual to whom users can feed back problems on behalf of a group or community of users may facilitate two-way communication between users and implementers.Our findings indicate that face-to-face contact is often preferred.It is also important that any such designated individual has the capacity to understand users' perspectives and their working environment.This does not necessarily require a person with a clinical background, but it needs someone capable of taking the boundary spanner role, who is able to 'translate' between management and users and build bridges between them.
We offer our model of user engagement in the context of national EHR implementations (Figure 2) as a starting point for conceptualising a complex phenomenon and emphasise the need to test and develop it further.It is important that future developments draw on relevant, existing models from other sectors as user engagement does not only present a problem in the context of healthcare IT.

Conclusions
Effective user engagement is critical for the successful implementation of EHR systems but extremely challenging to achieve, particularly in the context of attempting to implement a large-scale, complex and nationally procured system such as Lorenzo.In order to maximise the chances of success for similar initiatives internationally, it is important to allow local organisations to engage end users effectively, facilitate organic approaches to engagement and genuinely encourage and respond to user input at all stages of the implementation process.This includes focusing efforts on the initial 'point of engagement' by building on a common vision and by allowing user input into system choice.Mitigation strategies can be helpful but there is a need to recognise that these are often focused on reengagement.
Approaches need to shift from mechanistic models of clinical engagement-disengagement-re-engagement towards models that recognise the importance of the 'point of engagement', drivers for different groups of users, the need for continuous communication and key local individuals who are capable of being boundary spanners.
In relation to the implementation of EHRs in England, the new governmental direction including greater local systems choice and the resulting devolvement of engagement activities to individual hospitals is likely to facilitate user re-engagement. 60However, local efforts will need to focus on implementing software that is fit for purpose, which can be realised by user-informed design, and the coherent communication of the implementation strategy if users are to trust that their efforts to make the system work will eventually result in benefits.was chaired by David Bates.Other members of this committee included Martin Buxton, Antony Chuter, Ian Cowles and Kathy Mason.We also acknowledge the support of the National Institute for Health Research through the Comprehensive Clinical Research Network.

CONTRIBUTORS
AS conceived this study and KC led on writing the paper, with ZM, SC and AR collecting data, developing ideas and commenting on various drafts.AS and KC are guarantors.

Figure 1 A
Figure 1 A model of user engagement based on a literature review of technology implementation in the commercial sector (adapted from O'Brien and Toms 5 with permission of John Wiley and Sons) '... well I think if you're trying to promote change which is what the National Programme is all about, what informatics was set up to do, then I guess that if you're trying to convince someone to change you need credi-bility don't you?You need two types of credibility, one that what you're trying to do is compatible with their vision of the future and it's a good thing to do.And also you need them to feel that it's safe to change, you know, if I commit a change to my process to take advantage of all these systems are you going to support me?And that's what we're trying to work up.' (Interview, manager)

Table 1
A summary of our data set strategy (...) you know, shoe horn in something that isn't ready ...'(Interview, manager)

Table 2
Main themes identified, overlap with the existing literature and novel themes arising from the research User involvement in a national EHR implementation 197 essentially, obviously not personally but through his agents that we will be deploying this product in this many Trusts over this period of time and everybody on the [place] goes ''what?Nobody told us.''So then the CIOs [Chief Information Officers] had to go to their Trusts, you know, there's about seven acute Trusts and Mental Health Trusts and said you've got to deploy this product over this time period, then they would say to me what's the product and I'd say I don't know, there isn't one but there will be, trust me, you know, you can imagine what they would say to that.It's completely stupid, completely bonkers ...' (Interview, governmental stakeholder)