open space : The death rate for breast cancer has been falling for a decade, but nobody is clear why.Jo-Ann Mulligan and Anthony Harrison ask how the NHS and Whitehall can assess the value of healthcare in this field

What precisely do we get from healthcare in terms of health, compared to public spending on, say, education or social security benefits? Although the government is clearly committed to tackling those determinants of health which lie outside the health service, in reality its preoccupation is the NHS. This might explain why the emphasis on prevention and public health, so evident in the white paper, was not matched in the document that sets out how the extra money for health will be spent - the NHS plan.

But is it possible to disentangle the various influences on health improvement so as to allocate resources? What are the implications of not doing so? It is useful to look at a real example.

Death rates from breast cancer have been falling in the UK for the past decade or so. But no clear picture has yet emerged about how various developments in breast cancer interventions (screening, better diagnosis and new drug therapies such as tamoxifen) have contributed to this outcome. The evidence in relation to screening is particularly unclear, as the fall in mortality came too soon after the introduction of the national screening programme for all the benefits to be attributed to it. So it is all the more interesting to see what happened when The Lancet published a study by Peter Gotzsche and Ole Olsen last January which questioned the effectiveness of the programme. The press release from the Department of Health was swift and unequivocal: 'The NHS breast-screening programme is a success. . . There is no new evidence in [The Lancet's] report. . . The data has already been considered by expert committees. . . who found significant beneficial effects from screening.'

So That is all right then. The breast cancer screening programme is undoubtedly one of the DoH's more sacred cows. And the assumption that early detection leads to a more favourable prognosis has plenty of intuitive appeal. In fact most cancers are not identified through the formal screening programme, and some argue that many women present earlier because they are now generally better informed than 20 or so years ago.

Equally, however, the role played by drugs such as tamoxifen is often talked up. The news that mortality from breast cancer had dramatically fallen was greeted in the press last year with headlines like 'Breast cancer drug saves 20,000', even though the researchers themselves were clearly cautious in attributing the benefits.

Overall, it seems, neither the evidence presented by researchers nor the government response provide a clear picture of how much each of the various developments over recent years have contributed to the fall in breast cancer mortality.While it is perfectly possible that all have helped, unless the contributions of the different components are known there is a risk of double-counting the same benefits.

The inevitable row between screening enthusiasts and treatment advocates reflects how little we actually know about the way the cancer care system (and medical care more generally) is expected to improve health.We know a lot about the effects of specific interventions, but relatively little about how they interact with each other. Therefore, for those charged with organising services at a national, regional or local level, it is not clear how money should be allocated most usefully between services.

This vacuum in the knowledge base ought to be dealt with by the national cancer plan and the national service frameworks more generally.Yet none of these attempt to link specific proposals to the benefits that are hoped for. The cancer plan, for example, presents a long list of desirable activities, from prevention and screening through to treatment and investment in research, but does not identify what each is expected to achieve.

A real whole-systems approach to breast cancer (and other diseases) requires a more explicit model of how the healthcare system is expected to contribute to improved health. That model must acknowledge the influence of population-based characteristics (such as health-seeking behaviour) and the wider policy environment. This in turn requires a much better grasp of how the various factors interact with each other, in combination as well as separately. This would avoid double counting of the same benefits.When the cancer plan is eventually updated, we hope it will make explicit exactly how its proposals are expected to contribute to reduced mortality.

Jo-Ann Mulligan is research officer and Anthony Harrison is a fellow in the health systems programme, the King's Fund.