Stratification of patients by risk of acute hospitalisation – algorithm development, initial implementation and future strategy

For over a year Compass Health practices have been able to stratify their enrolled populations by predicted risk of acute hospitalisation. Alongside existing population profiling reports these risk scores aim to augment clinical judgement and assist practices in equitably targeting limited flexible funding to achieve maximum benefit.

This paper describes the risk algorithm, its development and implementation and also investigates the predictive accuracy of the algorithm, post implementation, by matching patient level predictions to actual hospitalisations. We present some representative cases of very high risk individuals, and provide summary statistics showing the shape of the risk distribution across the Compass population of around 300,000, and within specific practices.

Consideration is then given on how the algorithm could be further incorporated into future service developments around targeted programmes for patients with long term conditions.