It seems a deceptively simple plan - if you can identify the relatively small number of patients likely to use acute services intensively, you can concentrate on simpler, cheaper and more effective preventative care. It was a promise first held out in work by Kaiser Pemanente in the US and then imported by international healthcare companies such as UnitedHealth Europe.

It seems a deceptively simple plan - if you can identify the relatively small number of patients likely to use acute services intensively, you can concentrate on simpler, cheaper and more effective preventative care. It was a promise first held out in work by Kaiser Pemanente in the US and then imported by international healthcare companies such as UnitedHealth Europe.

But the indicators chosen to predict non-elective admissions remain controversial. For instance, an individual's admission history often proves an unreliable predictor of future care need - in effect, so-called frequent flyers do not remain so for long.

This is why the work going on in areas like Croydon and Norfolk is so important. It shows how the mathematics used to determine risk of multiple admissions has developed and been refined. The algorithm, developed by the King's Fund among others, is due for publication by the Department of Health next month and promises to help local clinicians and commissioners target intensive care at those most in need.

Croydon's work - which has been shortlisted in no less than four categories in this year's HSJAwards - involves the creation of 'virtual wards' to provide community support for the 0.3 per cent of the population at highest risk. Norfolk has adopted a different route, concentrating on helping people to avoid becoming high risk patients in the first place.

In both cases the projects are in their early stages and still working to win clinician support. However, the potential is great - both in terms of cash savings and in providing a service that can be measured not just in the volume of care that is heaped on an individual but the outcome.