The NHS has set a target to reduce emergency bed days by 5 per cent by 2008. Community matrons have been employed to help achieve this goal, yet they only target a very small number of people. This analysis shows how the service may be neglecting a much larger group who still have a significant risk of using secondary care.

These patients can be identified when we look at the pyramid-shaped chart, right, which divides the entire patient population according to risk of future hospital admission using a combination of inpatient, outpatient, accident and emergency and GP datasets.

The pyramid splits the population into four segments: very high risk of admission, high risk, medium risk and low risk. It demonstrates that case management is only capturing 0.5 per cent of the total at-risk population (each segment is labelled with an appropriate intervention strategy to cut admissions).

The bar chart shows the results of an analysis of a south London primary care trust's population. It reveals what percentage of overall service use is accounted for by the higher risk groups. It shows that if we just focus our efforts on case management (for the very high-risk groups) we can at most hope to influence 10 per cent of total emergency admissions.

However, if we also intervene to try to reduce the risk in the middle two segments (high and medium risk) which account for 19.5 per cent of the total population, we could then influence another 52 per cent of total emergency admissions.

Our analysis of the south London PCT's data shows the top three risk segments, or 20 per cent of the population, account for 62 per cent of emergency admissions.

Of course there is no reason why we cannot reach the remaining 38 per cent. However, because these admissions come from 80 per cent of the population, the appropriate intervention is likely to be population-based health and wellbeing strategies.

Chronic disease management therefore requires a whole population approach, not simply focusing on those at highest risk. Case management is important but, if we focus solely on it, we risk missing a large proportion of those at risk who would benefit from community-based interventions.

Stratifying the entire population by the risk of service utilisation allows us to provide tailored interventions according to patient risk.