Intended for healthcare professionals

Editorials

A new approach to weighted capitation

BMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6961.1031 (Published 22 October 1994) Cite this as: BMJ 1994;309:1031
  1. K Judge,
  2. N Mays

    Equity of access to health care on the basis of need alone is the central principle of the NHS. A corollary of this is that resources should be distributed among local health authorities in proportion to their relative health care needs. But this is difficult in practice because population needs cannot be measured directly and the proxies that are available are difficult to interpret.

    Since the publication of the report of the Resource Allocation Working Party in 1976,1 the allocation of resources to hospital and community health services has become progressively more complicated and controversial.2,3 The review of the Resource Allocation Working Party's formula in 1988, which resulted in the introduction of a weighted capitation formula, was widely criticised.4,5 Particular concern was expressed about how analyses of variations in hospital use among small areas were used to generate indicators of need.

    Prompted by the availability of new data from the census in 1991, the secretary of state for health announced a review of weighted capitation at the beginning of last year. Later this week ministers are expected to publish their conclusions. Although the basis for distributing hospital finances is likely to remain superficially familiar,6 substantial changes are expected.

    Three articles published in this week's journal explain the bulk of the analytical work that informed the review of weighted capitation. They use much more comprehensive data and more sophisticated statistical methods that are better informed by theory than hitherto. Consequently, they are an important contribution to thinking about needs assessment and allocation of resources.

    To begin with Carr-Hill et al describe an approach to a small area analysis of the determinants of hospital use, which was undertaken to identify population based indicators of the need for health care (p 1046).7 Statistical models of small area variations in use of hospital resources were developed in an attempt to distinguish between “legitimate” needs such as health status and deprivation and “illegitimate” influences such as the availability of hospitals beds, general practitioners, and nursing homes.

    In the second article Smith et al outline the main results of the analyses and show their possible impact at the level of the former regional health authorities (p 1050).8 Their four key findings are that the present method of weighted capitation overstates the costs of providing health care to elderly people; socioeconomic characteristics of areas are important determinants of the use of health care in addition to conventional indicators of health status such as mortality; small area indicators of need vary between different services such as acute and psychiatric specialities; and the availability of non-hospital resources has an important bearing on the demand for hospital care. The overall effect of taking these findings into account would be to redistribute resources to relatively deprived inner city areas at the expense of more affluent parts of the country.

    The final article, by Sheldon et al, describes an attempt to use routinely collected data about hospital use to identify population based indicators of need that improve the setting of budgets for general practice fundholders (p 1059).9 Despite using the most advanced statistical techniques the authors fail to find any convincing evidence that factors of need other than the age and sex structures of the local population are associated with small area variations in the use of hospital resources. They conclude with some justification that further research is needed to establish a resource allocation formula that is sensitive to variations in the costs of treatment. In the meantime continued vigilance will be required to ensure that incentives for general practitioners to exclude potentially expensive patients are not inadvertently created by any new method of allocating resources to fundholders.

    Despite, or even perhaps because of, their complexity the analyses presented in these papers represent a major advance in assessing the relative health care needs of small areas. They respond as far as possible to many of the common criticisms of previous attempts to develop empirically based resource allocation formulas. Nevertheless, some problems remain that are mainly due to inadequacies in the data available for analysis. One deficiency has been the lack of a consistent method of costing episodes of hospital use across specialties. Another is that the indicators used to measure access to health care and the need for treatment are highly correlated with each other, and this hugely complicates the statistical modelling. For example, whether the “illegitimate” influences of supply on observed patterns of hospital use have been, or indeed could ever be, fully taken into account remains debatable.

    A wider set of questions exists about the practical interpretation of the findings that relate to the narrow terms of reference within which the analysts had to work rather than the quality of their work itself. For example, although the review of weighted capitation was confined to the hospital and community health services, one may clearly infer from the papers that it is a serious mistake to evaluate relative needs for hospital services in isolation of an assessment of similar requirements for primary and community based care. A more comprehensive approach to establishing the relative health and social needs of different areas that transcends arbitrary service and administrative boundaries is now required. The latest research reported in this week's journal highlights the illogicality of, for example, allocating money to purchasers in London to buy hospital services without taking account of the relative lack of facilities for continuing care in the capital.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.