Research and Advances
Architecture and Hardware

Time to Rethink Health Care and ICT?

Success for a new U.K. national health care system depends on clinical context and scale, along with the capacity to emphasize interpersonal communication.
  1. Introduction
  2. Underpinning Concepts
  3. Conclusion
  4. References
  5. Authors
  6. Figures
  7. Tables

The health care sector has explored how information and communication technology might improve patient service for the past 50 years [8], but there is evidence that many, even most, health care information systems are failures [5]. Nonetheless, in the U.K., the National Health Service (NHS) has started to build a modern, dependable ICT infrastructure through an expenditure of £12–£20 billion over the next several years [9]. The unprecedented scale of the U.K. development, along with the scope and breadth of the NHS remit in providing universal cradle-to-grave health care for all U.K. subjects and the questions raised about the underlying models used in applying ICT to health care, suggests the U.K. experience has global applicability. We therefore explore the U.K. experience here as exemplar for our study of health care and ICT.

The National Programme for IT (NPfIT, run by the government agency NHS Connecting for Health) envisages a package of services across the U.K., including secure email, patient e-bookings, e-prescriptions, integrated care records, picture archiving and communications systems, a service for general practitioners, and a public health Web site, together with a national high-performance network [9]. It differs from other health care initiatives in two main ways:

  • The financial commitment exceeds anything done previously, especially in terms of capital commitment; and
  • The investment is at a national, rather than at the more conventional regional or even hospital level.

Given the stakes, it is worth exploring past experience, particularly in the U.K., to avoid repeating mistakes (see Figure 1).

The question of underinvestment is now being addressed in the U.K., but there is a long history of health care IS failures in the U.K. and elsewhere [1, 5, 7, 8]. Some of them were both high cost and high profile. Sinking money into health IS alone is not, it seems, the answer. The conventional critique also suggests a number of other issues, including the following:

  • Lack of fit between ICT applications and the work practices, environment, and culture they are expected to support;
  • Lack of robust, widely accepted evaluation methods, particularly with respect to cost;
  • Poor project management;
  • Inappropriate structure of the NHS;
  • Piecemeal nature of IS development since the early 1970s [1];
  • Organizational uncertainty; and
  • Pressure to roll out new ICT services before the pilots are fully evaluated as, for instance, with the Resource Management Initiative in the U.K. in the 1980s [5, 10].

Despite these criticisms, there is no apparent evidence that the NHS is any worse at buying and using IT than are the health services of other countries [1]. Indeed, as far as the primary care sector is concerned, a comparison in [3] of U.S. and U.K. health care claimed that the U.K. leads the world.

However, a major and fundamental issue has been largely ignored. It has been assumed, at least as evidenced by past projects, that health service applications are basically the same as any other application in business. This is not the case; health care is fundamentally different from other sectors, whether public or private.

Our argument has two perspectives. First, despite all the references to clinical applications and evaluations against particular user needs, health care IS has in practice adopted an “enterprise-type” model; that is, the application of ICT supports business functions (such as payroll, CRM, and ERP). In the “enterprise model,” communication is essentially the sharing of information supported through the IT infrastructure, rather than through person-to-person contact. Yet health care demands person-to-person interaction for collaborative diagnosis, treatment assessment, planning, and decision making. Such interaction must be supported to a much greater degree in health care than in other sectors, accentuating the need for interpersonal communication relative to the need for information services.

The second perspective is that the enterprise sets the scale of the model for IS in business sectors. Conventional IS is also designed to fit the inter-enterprise infrastructure, but the focus of the application is usually the enterprise itself or one of its subsets (such as factory, office, or headquarters building). Health care, on the other hand, is (or should be) a national-scale issue in the U.K. for which reason the focus should not be hospital, surgery, or clinic. This thinking is reflected mainly in the concept of integrated care pathways [2], cutting across the divide between primary and secondary care.

Attempts to build solutions around smaller units (such as hospital, surgery, and clinic), then connect them, even if they share a common infrastructure or services, following the inter-enterprise model, inevitably leads to suboptimal usage and communication barriers due to fragmentation of the infrastructure. The interesting U.K. development in this context is that the newly funded infrastructure is deliberately driven at a national level, a vision we view as positive.

The “enterprise” is too small a building block for health care, and models that start with national context, scale, and complexity might serve health care better.

To complete the picture, we note the following:

  • Secondary sector. In the secondary sector (hospitals), IS investment and attention has focused on management rather than on clinical issues, though pathology and, more recently, radiology have quickened the pace on the clinical front in the U.K. There needs to be much more holistic (systemic) thinking with regard to IT/IS provision; and
  • Primary sector. The primary sector has done better clinically (particularly in general practice surgeries), with electronic records and some decision support [3]. However, these applications have remained isolated. A major information exchange with hospitals involves patients who travel and the exchange of letters among doctors. This communication is not adequate for real-time care delivery.

We identify two notable examples of inadequate communication infrastructure among health care providers:

Regional Information Systems Plan (RISP). In the mid-1980s, the Wessex Regional Health Authority pursued this smaller-scale version of the current national vision to integrate health care IS. The ambitious aim at the time was to link every ward, surgery, and district nurse through the WRHA. The scale of the budget matched the scale of the vision. Unfortunately, it was also matched by the scale of mismanagement and purchasing failures [1]. For instance, the £3.3 million mainframe purchased for the project remained unused until its value had declined by over 75%. The failure to gain support from clinicians—in this case due to the system’s management rather than clinical focus—is a recurring theme in health care IS [1].

Resource Management Initiative (RMI). RMI, an initiative in the late 1980s, cost hundreds of millions of pounds to put IT systems into almost every U.K. hospital. This relative failure was analyzed in [2] in terms of the cultural problems of importing a U.S.-designed system—with its strong emphasis on cost recovery and specific clinical practices—into the U.K. environment. They note that less “than 30% of hospitals ever got their nursing information fully operational, and many of these were never fully used.”

Also worth asking is whether the system was ever likely to support the clinical side of care. An interesting approach would be to review the mission statements used by the different levels of management. The overall mandate was to enable the NHS to provide “a better service to its patients.” However, most local objectives focused on improving the services to their managers and clinicians [10]. It could be argued that RMI was not a failure at all, but that it subverted the top-level aspiration, which had a strong patient-centered element, in favor of more enterprise-oriented goals (such as managing patient throughput). In terms of helping managers, these managers were very keen on resource management [10]. Even clinicians favored it as a management tool; the approval rating (“strongly agree” or “agree”) was high; 68% of responding doctors and 76% of responding nurses felt it encouraged good working practice. The question of whether this would convert into better care was more mixed, especially for doctors, only 50% of whom approved the proposition that it would improve patient care. On the proposition that it provided care at lower cost, the approval rating of responding doctors was down to 34%.

In fact, there have been several high-profile failures, of which RISP and RMI are illustrative examples. Figure 1 sets them in that wider context, with our assessment of common elements of failure. The London Ambulance failure [1] echoes the failings of the RMI with, for instance, its poor fit with organizational culture and problems of management. Alarmingly, some failure signs around cultural fit have been evident in NPfIT since 2005 [6]. Clearly, cultural fit and clinical buy-in are long-term issues for ICT in health care.

Back to Top

Underpinning Concepts

While the examples we’ve cited are selective and the analysis incomplete, they do provide the scope needed to review the underpinning concepts about IS in health care. Otherwise, there is a danger of history repeating itself. Figure 2 points to questions related to health care IS by considering where the information resides or originates. You may disagree with our constituents of “data” and “information,” but many management processes and clinical pathways require the support of the kind of information access indicated by arrows A, B, C, and D in Figure 2; also see Table 1.

These information services and the processes they support have clear analogies in manufacturing, distribution, banking, retail, and other sectors as part of various enterprise models. This territory is well understood, though the enterprise level is suboptimal for the NHS as a whole. There is a range of interpersonal communication requirements unique to health care (arrow E in Figure 2). We define an “enterprise line” in general terms as a way to divide between the conventional services that may be adapted for health care from business and those that need fresh development for dedicated application to health care.

Yet even in these more traditional areas for IS, health care IS failure reflects, in part at least, a failure to do in health care what would be good practice in other sectors. Adapting good practice from other disciplines includes understanding the culture of the organization, capturing operational procedures, gaining buy-in of key stakeholders, specifying the changes in process expected to result from the new information infrastructure, and designing and managing accordingly. None of this way of thinking is new, and applying best practice in other sectors should improve the record of IS in the health sector.

Notwithstanding this omission, two elements of this model—above the enterprise line and below the enterprise line—require development in building health care information infrastructure over and above what is best practice elsewhere. With respect to the enterprise line in Figure 2, the more straightforward issue is above the line and the more complex one below.

Above the line. Is a national infrastructure simply a large example of the enterprise? We think not. The size and complexity of health care delivery systems and the breadth, scale, and sophistication of the services they offer make the two very different. The nature of the national infrastructure is a much more complicated proposition than even, for instance, multi-site manufacturing, where components and subassemblies are shipped from place to place along their journey. Table 2 identifies characteristics where health care might be similar or different from other sectors. It is clear that differences outweigh similarities and that these differences are crucial.

From a patient’s perspective, the problems of scale might involve having to coordinate the care of multiple pathologies through the general practitioner; for instance, the patient’s diabetes might require annual checkups at a local hospital; the patient’s cataract surgery might be a service offered through a treatment center; and the patient’s heart problems might relate to the diabetes but need to be treated through a tertiary referral center.

From a provider perspective, distributed patient care involves delivering highly integrated, personalized care requiring multidisciplinary teams located in different places. Coordination is a much more complex problem than, say, shipping and assembling components around a distributed supply chain.

These circumstances suggest that a more systemic model is required to better address the issues of scale and perspective. In the NPfIT initiative, the national scale has, for instance, been addressed, but a major issue concerning whether it will enjoy local uptake and improve clinical care remains. The early signs are not encouraging [6].

Below the line. The discussion in [11] began with the intriguing comment: “One of the challenges in the design of computer systems to assist health care providers is how to support collaboration while not requiring that people meet face-to-face.” But supporting this position is not enough; it is vital that support be given to below-the-line issues as well. Many communication issues (such as collaborative diagnosis of complex pathologies and bringing a specialist into a consultation between patient and physician) are below the line.

Support for below-the-line issues appears to be missing from much of the analysis within the U.K. health service. Face-to-face contact is fundamental to medicine and nursing. There is scope for the research needed to enable systems architects to understand the interpersonal dimension of care delivery (arrow E in Figure 2). Face-to-face contact can be between patient and doctor, patient and nurse, specialist and generalist, and specialist and specialist. It may concern joint diagnosis and planning or evaluation of treatment options. Again, this discussion is fundamentally different in type from that between, say, a production line worker wanting to speak directly to the production manager about a crisis on the line. They are complex and subtle yet essential to the effectiveness of health care IS.

Technologies are available for providing good face-to-face encounters using video communications, tracing the interaction, logging decisions, integrating the findings into care records, and more. However, little of them appears to have been embraced as a vital ingredient in models of health care IS, certainly as far as NPfIT is concerned at an operational or system design level. With regard to Table 3, the applications listed in the third column are being poorly implemented, if at all, because they are so different from what is available in other sectors.

It is not clear what the structure in Figure 2 might look like below the line. A careful taxonomy, followed by IS analysis, might provide a model that would greatly increase take up by clinicians and contribute to the quality and timeliness of care delivery. This is not a plea to give telemedicine a chance. Rather, it is encouragement to reexamine the role of interpersonal communication in health care IS models. This element is seriously underplayed in an enterprise-type understanding of the role of IS, rather than one that reflects the needs of patients.

Back to Top


There are many and varied reasons for the failure of health care IS, some relating to issues that are well understood in non-health care sectors of the economy. We recognize a common understanding of the role of IS in them in terms of an underpinning “enterprise-type” model. With the ongoing multi-billion-pound NPfIT deployment of information infrastructure in the U.K. it is important to explore this understanding in two areas: First, the “enterprise” is too small a building block for health care, and models that start with national context, scale, and complexity might serve health care better. This means that health service provision is different and should be looked at differently from any other industrial or government sector. Nevertheless, we recognize that lessons can still be learned from the business sector.

Second, better person-to-person models are needed to understand how the collegiate and interpersonal elements of care delivery could be embodied better in the systems used for care delivery. We have not sought to prescribe solutions but to encourage the IS community to critically consider existing models when addressing health care. This may not stop altogether the history of IS failure in health care, but once the more obvious failure mechanisms are addressed, clinical communities may be more positive about IS generally, making them more likely to benefit from its potential to help deliver the kind of service patients need most and win the public’s trust.

Back to Top

Back to Top

Back to Top


F1 Figure 1. Areas of concern in high-profile NHS IS projects.

F2 Figure 2. Health care applications classified based on where their information primarily resides.

Back to Top


T1 Table 1. Typical health care IS applications similar to those of any enterprise.

T2 Table 2. Characteristics of health care and other sectors compared.

T3 Table 3. IS applications in health care and other sectors compared.

Back to top

    1. Beynon-Davies, P. and Lloyd-Williams, M. When health information systems fail. Topics in Health Information Management 20, 1 (1999), 66–79.

    2. Campbell, H., Hotchkiss, R., Bradshaw, N., and Porteous, M. Integrated care pathways. British Medical Journal 316 (Jan. 10, 1998), 133–137.

    3. Detmer, D. Information technology for quality health care: A summary of United Kingdom and United States experiences. Quality in Health Care 9, 3 (2000), 181–189.

    4. Gray, J. and de Lusignan, S. National Electronic Library for Health. British Medical Journal 319 (Dec. 4, 1999), 1476–1479.

    5. Heeks, R., Mundy, D., and Salazar A. Why Health Care Information Systems Succeed or Fail. Manchester School of Environment and Development, Working Paper 9, Manchester, U.K., 1999;

    6. Hendy, J., Reeves, B., Fulop, N., Hutchings, A., and Masseria, C. Challenges to implementing the National Programme for Information Technology (NPfIT): A qualitative study. British Medical Journal 331 (Aug. 6, 2005), 331–336.

    7. Kaplan, B. Evaluating informatics applications—some alternative approaches: Theory, social interactionism, and a call for methodological pluralism. International Journal of Medical Informatics 64, 1 (2001), 39–56.

    8. Kaplan, B. The medical computing `lag': Perceptions of barriers to the application of computers to medicine. Int. J. Assess. Health Care 3, 1 (1987), 123–126.

    9. National Audit Office. Report by the Comptroller and Auditor General. Department of Health, the National Programme for IT in the NHS HC, Report No. HC1173, 2006;

    10. Packwood, T., Keen, J., and Buxton, M. Hospitals in Transition: The Resource Management Experiment. Open University Press, Milton Keynes, U.K., 1991.

    11. Safran, C., Jones, P., Rind, D., Bush, B., Cytryn, K., and Patel, V. Electronic communication and collaboration in a healthcare practice. Artificial Intelligence in Medicine 12, 2 (Feb. 1998), 137–151.

    12. Williams, J., Cheung, W., Cohen, D., Hutchings, H., Longo, M., and Russell, I. Can randomised trials rely on existing electronic data? A feasibility study to explore the value of routine data in health technology assessment. Health Technology Assessment 7, 26 (2003);

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More