An august collection of 200 European policy analysts, clinicians and health economists met recently in Helsinki to debate containing costs in their healthcare systems.

The discussion focused on hospital based treatments, and ignored almost every other aspect of cost containment, but the problems were identified clearly enough.

In any system based on 'premiums', be it home insurance or the NHS, the temptation is to overuse the service since it feels free; indeed, there is almost pressure to use it to get your money's worth. Insurance companies cover this with the carrot of a no claims bonus or the stick of more expensive premiums, but neither are available to traditional welfare systems so demand rises.

In addition, health is always a highly emotive issue, in which hope springs eternal and acceptance becomes ever more difficult, meaning perverse pressures arise. Rationality tends to disappear when our health and mortality are at stake, and the so called 'rescue principle' kicks in even when logic says no more can be done.

This is what makes us spend vast fortunes 'striving officiously' to keep untreatable patients alive. But it keeps our services humane and human.

Now throw in the fact that medicine is an inexact science, in which cause and effect are hard to link, especially in chronic illness, primary and community care. The temptations of multiple complementary therapies become clearer, especially if the free system is paying.

Helsinki's approach was to work out ever more complex methods of handling waiting lists or defining booked admission systems that seemed (perhaps deliberately) to miss the point: one cannot discuss cost containment without demand containment. This is because with increasing demand for any service, there are three possible solutions: spend more, be more efficient, or restrict the supply of services.

Spending more does not seem to be an option anywhere in Europe, despite all the rhetoric about proportions of GDP spent on healthcare (boring to everybody but the Brits).

Increasing efficiency is a tempting route since it seems to promise more for less. If we are going to look at efficiency, we must look at the whole system, not just single parts. The effects on the whole welfare state of an intervention in one sector alone are vast; increasing throughput for elective surgery, for example, puts pressure on primary and social care; restricting money to social service departments creates an expensive backlog of bed blocking.

So we should measure efficiency gains in appropriate terms across the whole system. That raises questions about the extent of the system. Do we include local authority services?

Social security? The costs of informal care? Don't ignore reward mechanisms by looking only at treatments. Look at remuneration, peer approval and all the other incentives that make self-driven, motivated professionals work more effectively.

Look, too, at the locus of treatment and all the professionals - nurses and pharmacists as well as doctors. Most NHS expenditure is on staff, and most capital is tied up in enormous hospitals whose impact on the population's health is marginal at best.

The third option is to restrict supply. Once again, this needs to be considered across the whole system, which must include the public. As long as demand exceeds supply and the current value system continues to drive up expectation, service reconfiguration and substitution will not reduce costs; at best they may increase throughput. In the end, we will have to exercise some form of overt restriction. No matter how efficiently we spend our health serv ice euro, we w i l l need to f ind an equitable, fair and kindly way of rationing the supply of public services.

So what about the following?

There are basically three types of health service intervention: those carried out across a population (public health medicine); the essentially technical (evidence-based 'cure', including surgery and proven drug therapies), and areas of 'care' which, in Voltaire's words, 'entertain the patient while nature takes its course'. This last is driven substantially by personal values, and predominates in those parts of medicine where cause and effect are harder to determine.

Since population-wide initiatives are hard to carry out locally, it makes sense for the state to provide and pay for them from taxes. Technical procedures are relatively easy to quantify, so they could be paid for from some form of 'insurance'-based model. This could be taxation-based (after all, the NHS is just a large insurance scheme, sharing costs and risks across an enormous client base) or paid for through, say, employment-based insurance schemes.

With public health and 'cure' sorted, we are left with the 'care' niche. For this, we could have some form of compulsory insurance but no standardised provision. In that way, people could access the service that suits them.

The 'doctor as treatment' is an important part of any healthcare system, even in the developed world, and allowing people to work within their own belief systems might be an excellent way of maximising the placebo effect we have ignored for so long.

The model needs some more work, of course; cure is likely to subsume care, but then that's exactly what's happening now, as general practice becomes more technical, and patients find support and succour in the homeopathic clinic or iridologist's office.

What would happen to long-term care, innovation and governance? And (back to Helsinki) somebody would still need to make the choices about which technical services to provide.

We really can't escape demand management, can we?

Jonathan Shapiro is a senior fellow at Birmingham University's health services management centre.