INTRODUCTION

“We seek the development of a learning health system that is designed to generate and apply the best evidence for the collaborative healthcare choices of each patient and provider; to drive the process of discovery as a natural outgrowth of patient care; and to ensure innovation, quality, safety, and value in health care”.

Harvey V. Fineberg, President, Institute of Medicine1

Healthcare is becoming progressively more complex, with an ever increasing array of treatment options, older and frailer patient populations and increasing costs.2^,3 Creating learning health systems is now widely seen as essential to address these challenges through strategically collecting electronic data and repeatedly using these to improve clinical and organisational processes.4

Hospital electronic prescribing and medicines administration (HEPMA) systems, which allow electronic management of prescribing and administration activities as well as analysis of data collected,5 have of late received considerable policy interest in Scotland and the United Kingdom (UK). This is in the main being fuelled by the desire to improve prescribing safety in hospitals.6 What is, however, often less well appreciated is that proactive reuse of HEPMA-generated data can also bring considerable benefits. Whilst this is to an extent true of all health information technology (HIT), the planned implementation of HEPMA systems at scale across National Health Service (NHS) Scotland and the fact these systems have the potential to transcend professional and organisational boundaries provided us with an important opportunity and lens through which to investigate how best to achieve data-informed service delivery and decision making in the NHS.7^–9

Creating learning health systems through effective use and reuse of electronic data is extremely challenging. Underlying reasons are both technical and social in nature.10^,11 Social aspects may, amongst others, include ongoing problems with system usability and integration with clinical workflows, which can contribute to poor quality data being entered, which fails to reflect actual clinical problems and/or procedures. Other factors may include social structures and relevant associated actors such as policy makers, system suppliers, professional groups, organisations, legal entities and patient groups. These can have important effects through, for instance, setting standards and providing guidance, making and influencing strategic decisions, and through direct implementation-related activities. Wider technical issues may include a lack of system interoperability and associated inhibited meaningful exchange and potential for aggregation of data. An integrated overall data strategy may help to mitigate these risks, ensuring that social and technical systems are set up from the start to realise effective use and reuse of data to improve quality and efficiency.

Many international examples exist of how such large-scale data strategies have been developed in an ad hoc manner, often rather reactively addressing problems as they emerge as opposed to strategically planning for the efficient reuse of data from the start.12^,13 Particularly, little is known about the needs of different stakeholders in relation to effective use and reuse of health-related data, although the value of multi-stakeholder perspectives has been repeatedly advocated in informing policy and some valuable insights have been generated in international settings.14^–16 However, these may not be applicable to the Scottish context. Consequently, there is a danger of misalignment between needs and healthcare strategy. In addition, there is continuing global debate surrounding the degree of central guidance needed and warranted to support such initiatives.17 NHS Scotland is in a unique position in this respect as the country is relatively small (allowing a degree of centralised guidance that is simply not feasible in larger countries; see Box 1) and some specialty areas have already made significant progress in collecting and reusing data at scale.

Given the opportunity offered by the planned national implementation of HEPMA systems across NHS Scotland, we sought to investigate stakeholder perspectives and experiences surrounding the continued implementation of HEPMA systems and how these can inform political strategy surrounding the use and reuse of the digitised data generated through HEPMA. HEPMA systems were chosen as a timely exemplar to explore issues related to wider health information infrastructures and associated strategy to create learning health systems.18^,19

RESULTS

Overall, we invited 36 individuals, of whom 21 attended on the day (Box 3). The others declined due to other prior commitments or did not reply to email contact. We summarise key emerging themes in Box 4. A range of stakeholder needs and resulting implications for policy emerged from this work. Throughout, it became apparent that diverse stakeholder groups had different priorities. For instance, whilst healthcare professionals were mainly concerned with system usability and appropriate resourcing, system suppliers and policymakers were primarily interested in exploring issues relating to data reuse at scale.

DISCUSSION

The identification of different stakeholder experiences and needs is important to catalyse the development of an integrated data strategy to support the effective use and reuse of HEPMA data. Adding to the existing international implementation-related literature, we have identified a number of micro-, meso- and macro-factors that are important in facilitating this in the Scottish context. In particular, our work has highlighted that appropriate user interfaces and sufficient resources are needed to implement systems, and that appropriate infrastructures need to be in place to support collecting data for reuse. Such infrastructural considerations are, however, often neglected in international discourses surrounding data strategies, particularly those surrounding capacity development.14^–16 Our work further indicates that throughout data strategy planning, it is important to aim for the creation of a learning health system at individual healthcare worker and organisational levels, as this can facilitate data collection and analysis.30 In doing so, there is a need to ensure that users and organisations are motivated and supported to collect accurate data, e.g. in order to improve clinical and organisational performance. Although motivating clinical users has been the subject of many implementation-related efforts, devising effective ways to feedback clinically relevant data to individual users is still a subject of much debate.31 Examining motivations of organisations in reporting data for national aggregation beyond financial incentives is an other issue that has received very little attention and would present a useful next step in exploring how to promote the adoption of national data strategies.

This work may be seen as a starting point, as it has begun investigating how policy decisions in relation to developing strategy to create learning health systems in NHS Scotland may be conceptualised. As such, we have provided contemporaneous insights into senior stakeholder viewpoints associated with the key components of a national strategy, illustrated the main concerns from a variety of perspectives, and provided implications of this work for policy development and future evaluation activities.32 These findings now need to be incorporated into ongoing UK-wide initiatives such as the Farr Institute of Health Informatics Research,33 which is aiming to promote research through data linkage. However, although we facilitated interactive discussions involving a broad range of stakeholders, this work was necessarily limited in scope as only a small number of senior stakeholders were participating (otherwise, the interactive nature of discussions would have been compromised). The transferability of this work to other UK and international contexts may therefore need to be assessed by feeding back emerging results to a wider range of stakeholders and refining policy accordingly. In addition, the close involvement of policy makers may have influenced the results.

A number of recommendations for national strategies to facilitate the creation of learning health systems through HEPMA systems emerge from this work, which we have summarised in Box 5. Although these are likely to be more relevant to countries beginning the journey of system procurement (such as Scotland), they may also be applicable to countries considering a more centralised data strategy in order to effectively aggregate data from locally implemented systems.

Establishing robust information infrastructures is a key for any national healthcare strategy. With the aim of aggregating data on larger and larger scales, such infrastructures may be comparable to other large information networks carrying electronic information like the internet and multi-national corporate systems.29^,34 There is therefore a wealth of experience with similar technologies from the information systems literature that healthcare strategy can draw on, particularly in relation to developing risk mitigation strategies. For instance, it is important to recognise that such systems are evolving over time and that they have both technological and social elements.29 In our work, these related to technological systems (hardware and software to collect data) as well as users and organisations (the social element that involves collection and analysis of data). As a result, any strategy to implement associated technologies and social processes needs to be flexible enough to cope with the evolving nature of such systems and their evolving use over time.34 Similarly, technologies need to be adaptable enough to allow for changes in strategy (e.g. in data to be collected) and social contexts of use (e.g. changing professional structures). A number of implications for the design of systems drawing on the literature surrounding information infrastructures have already been outlined and may be applicable to healthcare. These include the need to ensure usability, if possible draw on existing infrastructures, expand use of systems with persuasive tactics, make technological systems easy to use to promote usability and modularise the system in order to promote flexibility.29

International experiences may also facilitate the refinement of a national (Scotland- or UK-wide) healthcare strategy. For instance, ongoing work surrounding digital maturity in England could help other countries to benchmark and serve as a starting point to develop relevant guidance.35 Similarly, experiences from large health systems in North American settings may help to develop governance structures (such as, for example, surrounding data stewardship and potential partnerships with industry), and promote learning from experiences (e.g. surrounding dealing with potential confounders and timeliness of data, or in helping to establish specialist roles that emerge from new systems such as HEPMA Pharmacists).36 Drawing on experiences from other industries, such as banking and retail, may also be fruitful in terms of future work. There will further be a need to ensure continued national high-level leadership and drive through, for instance, the appointment of Chief Data Scientists,37 and the continued use of financial incentives to promote collection of electronic health data.38

CONCLUSIONS

Achieving the vision of learning health systems can be greatly facilitated by drawing on existing national and international experiences. The example of developing HEPMA strategy within NHS Scotland has allowed us to identify key considerations that should be taken into account when planning for national strategies to aggregate healthcare data. These include defining minimum datasets, creating incentives, and promoting a learning health system through data feedback loops at different levels. At present, a truly learning health system is still a vision in most countries, but this work may be used as a starting point to promote international efforts to aggregate data at national scales.