PCTs to receive predictive risk software

Software to help identify patients at high risk of readmission to hospital is to be sent out to all primary care trusts next month.

The program consists of an algorithm that PCTs can apply to hospital episode statistics (HES) to identify high risk patients with long term conditions. It will take account of the number of previous admissions a patient has had together with data on diagnosis, medications, demography as well as the relative rate of individual hospitals for subsequent readmissions.

The risk prediction system has been commissioned by the Modernisation Agency and all 28 SHAs in England to find a more robust way to classify PCT populations according to potential risk.

By identifying high risk patients, SHAs hope that PCTs will then be able to target better ‘upstream’ care or case management to help keep such patients out of hospital using community matron .

Kate Gill, project director for long term conditions at Essex SHA which is leading the commissioning process, said the software aimed to help PCTs maximise the effectiveness of their interventions.

The project was commissioned because previous methods of identifying high risk patients either through counting the number of previous admissions or based on clinical knowledge have proven to be inaccurate.

Gill told EHI Primary Care: “We will be doing a formal launch in the middle of August when the software will be going out to every PCT in England.”

The project is being run by the King’s Fund along with New York University and Health Dialog Data Service, a US health data analysis company. The project, which began in April, is in three phases consisting of a literature review which has now been completed [Word, 523K] the first version of the algorithm which will be sent out next month and a further version of the algorithm which will be completed by the end of the year.

The second version of the algorithm aims to link HES data from community services such as GP records, district nursing records and social services data.