Intended for healthcare professionals

Education And Debate The hospital of the future

Planning hospitals with limited evidence: a research and policy problem

BMJ 1999; 319 doi: (Published 20 November 1999) Cite this as: BMJ 1999;319:1361
  1. Nigel Edwards, policy director (nigel.edwards{at},
  2. Anthony Harrison, fellow in health policy analysisb
  1. a NHS Confederation, London SW1P 4ND
  2. b King's Fund, London W1M 0AN
  1. Correspondence to: N Edwards

    This is the last in a series of seven articles

    Research into the running and planning of hospital services has been neglected. This is surprising given the importance of hospitals for the public, politicians, and the healthcare system (box 1).

    Box 1: The importance of hospitals

    • Hospitals account for 40-60% of health expenditure in the Organisation for Economic Cooperation and Development and over 70% in the former Soviet Union and eastern Europe

    • Hospitals have been subject to major changes—for example, in the United Kingdom bed numbers were reduced by 15.5% between 1987-8 and 1997-8. Hospitals are important symbols of civic pride1

    • Hospitals make a significant contribution to education and employment. In London, for example, it is calculated that the medical schools and their associated hospitals contribute up to £500m to the local economy2

    • In the United Kingdom, 20% of the population are admitted annually and over 25% use an accident and emergency department


    The articles in this series have shown that despite this importance there is a paucity of research and policy about how hospitals work, how they should be staffed, what size they should be, and how change can be managed. This article examines the main gaps in our knowledge about the future of the hospital sector identified in this series and examines the implications of this for policy and research (box 2).

    Summary points

    Hospital planning is done on the basis of limited research

    There is little evaluation of completed plans

    Many of the assumptions used are not stated clearly and are often based on limited or poor evidence—this applies to many of the arguments for increased centralisation

    The paradox of increasing admissions and falling bed numbers has contributed to the problems of responding to emergency care

    Planning needs to take into account the limited state of knowledge

    Box 2: Gaps in research

    There are many uncertainties about the future of hospitals. Some of the more important areas where research is needed are:

    • Understanding likely future demand

    • Improved forecasting of length of stay and other aspects of hospital performance

    • The benefits of subspecialisation and methods to reduce the impact of this on access to services

    • How hospitals can be staffed in new and imaginative ways

    • The development of flexible approaches to planning hospitals

    • How to manage change effectively


    The hospital

    Analysis of trust business cases for rebuilding and other developments and strategies proposed by health authorities shows wide variations in the assumptions made about almost every aspect of future hospital provision. Despite the apparently increasing difficulty of meeting growing demand for hospital care many hospital plans envisage substantial reductions in the number of beds, and hospitals with large private finance initiative schemes expect reductions of 20-30%.3 But whether these can be justified in terms of either future demand or levels of performance is unclear.

    Demand for hospital care

    The demand for hospital care is poorly understood and underresearched. Admission rates grew by 40% between 1988-9 and 1997-8, and only part of this is explained by changes in population or morbidity. The remaining growth is usually attributed to new technology and poorly explained exogenous factors such as changes in the attitudes of patients or general practitioners, or in the way elderly people are treated.

    Planners differ widely in their assumptions about future growth in demand. Some plans assume increases in demand in excess of those suggested by population growth, others ignore these exogenous factors, and in a few cases assumptions have been made that demand may be reduced. Planners are clearly making assumptions about changes in morbidity or success in managing demand, but these are rarely stated explicitly or supported by evidence.

    Performance of hospitals in future

    The assumptions made about future length of stay show similarly wide variations where benchmarking and discussions with local clinicians are often used to develop forecasts. Since linear forecasting produces absurd results (such as negative length of stay beyond 20294) and the downward trend in the number of beds has flattened, these forecasts increasingly rely on assumptions about the use of hospital at home, nursing homes, and other forms of “intermediate” care. There is reasonable evidence for the clinical effectiveness of these interventions, but more evidence is required about the total number of patients who may be eligible, whether the costs of these services are less than those of hospital, and the extent to which these services can be used as the basis for ambitious assumptions about the future use of beds.

    In addition to quite different forecasts of day case activity many plans seem to make the assumption that growth in day cases will represent a direct substitution for inpatient work However, the evidence of the past 15 years is that much of the growth in day case and outpatient procedures is the result of new techniques and technology.5

    As for planning levels of bed occupancy, there is a failure to appreciate that planning for a mean occupancy of 90% guarantees that hospitals will have insufficient numbers of beds on a substantial number of occasions because of the inevitable variations in daily admissions Furthermore, it is often not appreciated that reductions in length of stay often require lower occupancy rates to retain sufficient flexibility to deal with random fluctuations in demand.

    Difficulties in forecasting hospital requirements are matched by difficulties in managing hospitals and balancing the conflicting pressures of emergencies and the requirement to reduce waiting lists. Virtually no research has been done on the best way of handling these conflicting pressures on a day to day basis, or on the cost or clinical implications of achieving greater separation in organisational and physical terms, either within a given hospital or through the creation of separate, stand alone institutions.

    The concern about simultaneous and apparently contradictory reductions in the number of beds and increases in the number of admissions has led to the establishment of a review of bed numbers by the secretary of state. International experience, however, shows that setting standard assumptions has major hazards, the most dangerous of which is to institutionalise the current state of health care. Just as dangerous is the view that simple calculations can predict need and convert it into future provision. On the contrary, uncertainty about future requirements cannot be eliminated. Therefore, cost effective means of building in physical and operational flexibility must be found, and this has also been neglected.

    Changes in the way that hospitals provide care have implications for clinical training and for the working environment of clinical staff as well as for future staffing requirements. Because responsibility for these different areas is divided among the professions, training organisations, the Department of Health, and the NHS Executive, they have been persistently neglected.

    Hospital systems

    The development of the internal market and creation of trusts has produced incentives for hospitals to plan on the basis of maximising the role and status of the individual trust. We have, however, observed over the past two years an increasing trend for groups of hospitals to work more collaboratively, and a softening, or even abandonment, of the competitive ethos, in line with the government's white paper The New NHS.6

    The result of this is that the type and range of options that are considered to be available change when the objective is how to plan for an area where a number of hospitals form a potential network of complementary provision, rather than how to compete. Joint plans are increasingly likely to be followed by mergers, the ultimate surrender of individual aspirations to the collective will. The recent acute strategy for Scotland is an early example of what seems to be a growing trend towards planning on a system-wide basis.

    Embedded Image

    St Thomas's Hospital, London, in 1871


    This development reflects the view that some services must be organised on a scale larger than any one hospital, for some services for populations as large as one million. However, as the Calman Hine report recognised for cancer care, the amount of evidence bearing on such large scale issues is limited.7 Furthermore, neither the Calman Hine report nor the Scottish strategy report8 adequately deals with the relation between services organised in this way, as they do not allow for the impact of their proposals on the way other services are provided—even though the same staff and facilities may be involved.

    The wider system

    Both the demands placed on hospitals and their efficiency as providers depend on the nature and effectiveness of community based services such as rehabilitation facilities and out of hours cooperatives formed by general practitioners. They also depend on the way in which potential users, particularly of emergency services, decide whether and how to access care. Although the phrase “whole systems approach” has now found its way into official documents, virtually no research has been commissioned at the “whole system” level.

    Staffing and medical specialisation

    One of the most powerful factors making for change in hospitals has been increasing medical specialisation. As the recent review by the York Centre for Reviews and Dissemination9 and Posnett's article in this series10 have shown, high quality evidence on the benefits of this process of centralisation and specialisation is limited. Nevertheless, the recent recommendations for hospitals by the BMA and the Royal Colleges of Physicians and Surgeons envisage a continuation of this process.11 Furthermore, the colleges are issuing guidance that will put managers in a position in which they will have to close or reduce the role of some hospitals.

    This is in direct opposition to the high value placed on access by the public, and unless models can be developed to overcome this it is possible that the accountability of the colleges will be questioned, and they may come into direct conflict with politicians. The problem is particularly acute in more rural areas, where even the revised minimum populations suggested in the most recent document by the BMA and the royal colleges may be hard to achieve. A compromise will need to be developed between the requirements of education and training and the development of local services, and some hospitals may not be able to continue to function as educational establishments responsible for training junior doctors.

    Changes in the way that hospitals provide care have implications for clinical training and the working environment of clinical staff as well as for future staffing requirements, and vice versa. Because responsibility for these different areas is divided among the professions, training organisations, the Department of Health, and the NHS Executive, the links between them have been persistently neglected.

    The results of this are apparent in the current crisis in the recruitment of nursing staff. Although many other factors play a part, one element is the lack of research on the number and type of nurses that hospitals require and the contribution of nursing to patient outcomes.

    The research agenda

    The previous sections have focused on the areas where we believe that more research is required. Two general points need to be made.

    Firstly, research relevant to hospitals has tended to concentrate on single interventions and less frequently on some models of service delivery such as hospital at home. It tends not to address issues about the planning of whole systems, and it is rare for the results of hospital reconfiguration to be evaluated. Although there is a requirement for large capital schemes to be evaluated after completion, this does not seem to happen routinely, and where such evaluations are carried out the results are often not in the public domain. Politicians and policymakers may find the critical evaluation of previous decisions uncomfortable, but unless it is carried out mistakes will be repeated and there will be no collective learning within the NHS about how to plan such schemes.

    In the past, funding to support this type of research has been limited. The new research into service delivery and organisation to be commissioned by the NHS Executive offers the potential for many of these areas to be investigated.

    Even though this initiative is welcome, it will not be enough. An additional problem is that the time lag in implementation means that evaluation may become history rather than research and, given the pace of change, the past may not be a reliable guide to the future. No substantial sources of funding have been available to support this type of research. Moreover, many of the questions for which planners, managers, and policymakers need answers are not easily answered with traditional methods of research into the health service.

    Secondly, new research techniques are required to support planning for the hospitals of the future, including the development of scenario planning and modelling, and simulation techniques to identify uncertainties and the sensitivity of plans to forecasting errors There should be more evaluation of completed plans and much better systems to exchange knowledge about innovations. Some nationally led experiments are also needed—in the development of service models—for example, for rural areas where the trends referred to above are undermining existing patterns of provision.


    Little research has been done that highlights the central issues of hospital planning: how many hospitals we need, what services each should offer, how they should relate to each other, and how, once these issues are resolved, they should be organised, staffed, and managed.

    Plans for the future of hospitals need to recognise our lack of knowledge, and, if there is to be central guidance, this should be that whatever is planned should be robust in as many possible futures as are conceivable. Research should be directed at understanding how flexibility can be incorporated into hospital design at low cost. In this respect there is perhaps some reason to be concerned about the impact of the private finance initiative. These schemes will have fewer beds but may not incorporate design ideas that allow flexibility since, in many cases, the costs associated with the planning will fall on the NHS.

    This failure of research reflects a larger failure to take the planning of hospitals seriously, which has been particularly marked since regional health authorities were abolished. Although geographical variations rule out a “one size fits all” approach, we have identified a range of issues that require a central response. The professions have begun with the publication of a consultation document on acute hospital services to respond to this challenge.12 We can only hope that the Department of Health will do so too.


    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract