In 2002, the National Health Service (NHS) in England and Wales planned to invest over £2 billion in information and communication technology (ICT) [1]. This includes initiatives such as electronic patient records, electronic prescribing, the NHS Direct Telephone and Internet Service, and the National Electronic Library for Health [1,2]. Researchers from multiple disciplines in the UK and elsewhere had been investigating health informatics, but the NHS, despite its large investment in ICT, had not set a national research agenda for ICT.

A one-day conference in 2002 on Health Informatics Research and Development, sponsored by the research councils, Department of Health, and Department of Trade and Industry, concluded that the “lack of national strategy, capacity and career paths in health informatics have been weaknesses and remain threats to realising the informatics potential of the National Health Service…. [L]arge investment in the National Health Service and e-Science is unlikely to achieve its objectives without radical improvement in support for academic health informatics…. [This emphasizes the] importance of...clarifying the academic agenda for health informatics” [3]. The short- and medium-term challenges were seen as the following: (1) establishing the foundations of a knowledge infrastructure, (2) innovations in the clinician computer interface, (3) workable privacy protection, (4) more creation of knowledge from routinely collected data, and (5) finding the metrics of success for health informatics.

The NHS has three main national NHS research and development programs: Health Technology Assessment, New and Emerging Applications of Technology, and Service Delivery and Organisation (SDO). The SDO was launched on March 30, 2000, to consolidate and develop the evidence base on the organization, management, and delivery of health care services [4]. To respond to the needs of the “stakeholders” (service users, health professionals, and policy makers), the NHS, through its SDO research program, undertook an initial “listening exercise”[5] to produce a document outlining its overall priorities for research. It has continued to use this approach to develop and commission research [6]. It commissions a “scoping exercise” (normally a literature review and a stakeholder consultation) and then uses that in subsequent calls for proposals.

This study explored the concerns of professional and lay stakeholders regarding future developments of eHealth and reviewed relevant policy to produce recommendations for eHealth research. A parallel literature review was carried out by others and has been reported elsewhere [7-9].

In the context of the British health service, which is mostly free at the point of delivery, the overriding concern of stakeholders was that spending money on eHealth should be worthwhile and should lead to improved health and quality of life. At first this appears an unremarkable finding. However, although such an aim is part of the political rhetoric and may be an unstated assumption of policy documents in the United Kingdom, it is not often explicitly addressed in service development and use of information and communication technology. It is significant, for example, that the NHS recently funded a program originally called the National Programme for Information Technology, rather than a program for eHealth. (Subsequently, the program has been renamed Connecting for Health).

On the other hand, we know that doing research that can show a difference in health as a result of an eHealth intervention is difficult, partly because partial implementation of an e-booking system or a hospital information system is difficult. Gold standard randomized trials whose results can be generalized for widespread implementation are very difficult. In addition, to see changes in health or to measure cost benefit is slow and expensive, made more so now in the United Kingdom by the time needed for increased levels of ethical and research control and approval [12,13]. Thirty years ago, Blum noted that half the papers about computer applications concerned systems that were no longer operational [14]. We all know that ICT changes become ever more rapid. Research, therefore, has to be more about human behavior and how we use information and less about specific organizational or technological environments. It is essential to recognize the difficulties of addressing stakeholders’ needs by measuring change in health outcomes.

In some cases, modelling the longer-term costs and benefits (in terms of health) may be desirable. As systems continue to evolve, the health benefits may be seen not in the immediate change, but in a future evolution made possible by the initial change [15-17].

Many of the recommendations derived from both the stakeholder consultation and policy review confirmed the need to identify best practice and the barriers to implementation of that best practice. There are many examples of medical informatics research with demonstrable benefit which decades later still wait to be implemented more widely. For example, computers have been used successfully for patient interviews for nearly 30 years. Slack et al first reported on a computer-based medical history system in the New England Journal of Medicine in 1966 [18]. Yet, despite numerous research reports [19], the method has not been routinely adopted.

Rogers’ description of the diffusion of innovations [20] is well known, but stakeholders and policy makers want to see this process accelerated. The recommendations included, for example, identifying and exploring examples of effective informatics training for health professionals to see how these can be disseminated, or investigating working practices in other sectors (eg, e-business) to identify best practice and barriers to similar uses of ICT in the health sector.

In discussion, professional stakeholders often reverted to their role as patient or consumer rather than, for example, speaking as a supplier of ICT or from the point of view of primary care. Thus, although we sampled from all segments of the stakeholder population, there were no obvious differences between the different types of stakeholder. (Our sample was small, however, making our power to detect differences limited.)

The four categories which emerged as the grouped themes from the data are similar to classifications and descriptions found in textbooks of health informatics (eg, [21]). Blum, in a personal review of Medical Informatics in the United States, 1950-1975, presents a historical table of the Scope of Medical Computing from 1950 to 1980 using the three headings Data, Information, and Knowledge Applications [14]. He said, “Within a category, research begins only after the supporting technology is mature enough to support it beyond the conceptual level.” His table showed that he viewed data applications in the 1980s as refined, information applications as mature, and knowledge applications as prototype.

Four limitations of our study are the following: (1) We were not able to recruit people who were very sceptical about the potential of eHealth. However, we think that a more sceptical sample would be likely to have expressed similar concerns about improved health and value for money. (2) We did not have equal representation for the 12 predefined groups. However, we achieved coverage (although, in some cases, only one member; see Table 1) for all groups. Furthermore, as most participants often “reverted” to their role as patient in the discussion, our original idea that different professional groups might have certain biases and try to put forward ideas to their advantage seemed wrong (although numbers were small to detect any differences between groups). (3) Our lay sample was recruited only in the South East and the South West of England and did not include any people from ethnic minorities. We have no knowledge or hypotheses about how views may differ in groups not represented. (4) An eHealth agenda derived from policy review and stakeholder consultation in a country with a state-run health service may not transfer well to countries with private or insurance-based systems.

The purpose of our study was to produce a conceptual map and research agenda for eHealth based on stakeholder views and policy review. In a parallel SDO project, Pagliari et al [7-9] developed recommendations based on a review of existing academic and wider evidence sources, indicating the scope of the eHealth concept, the effectiveness of eHealth innovations, issues for implementation, and future directions for eHealth. Their research evidence is grouped by four broad technological categories:

Their results are also interpreted with respect to the following broad content areas:

While differences in emphasis were expected due to the methodology of each study, the clear parallels between the results offer support for our recommendations. Our conceptual map, which has come from stakeholder discussion and policy review, also helps to put both our own and other detailed recommendations into a framework concerned with information and how we use it.