parr case finding tool patients at risk of re- hospitalisation
TRANSCRIPT
PARR case finding tool
Patients at risk of re-hospitalisation
Background
Risk prediction system for use by PCTs
Identifies patients at high risk of emergency re-admission to hospital
System produced by Kings Fund, New York University and Health dialog data service
Commissioned by Essex SHA on behalf of the 28 SHAs
It’s FREE!!!!
Background to project
Phase 1 – Literature review: June 2005
Phase 2 – Development of an algorithm that uses HES data to predict future risks: July 2005
Phase 3 – Development of an algorithm that links HES with other routine data on utilisation of care, in order to predict risks: January 2006
The PARR case finding algorithm
Uses hospital admission data to identify patients at high risk of re-hospitalisation in the 12 months following a “reference” hospitalisation
Produces a “risk score” for probability of future admissions which draws upon broad range of information about the patient – current hospitalisation, past hospitalisation, geographic area where patient resides, hospital of current admission
Risk scores range from 1 to 100 – higher scores having a higher risk of admission in next 12 months
Output……
PARR risk score % flagged patients admitted within 12 months
0 - 10 0.0%
11 - 20 11.0%
21 - 30 23.8%
31 - 40 35.1%
41 - 50 45.3%
51 - 60 56.4%
61 - 70 66.0%
71 - 80 73.6%
81 - 90 80.5%
91 - 100 91.0%
Characteristics of patients flagged with high risk scores (over 50): Higher level of utilisation
Significantly older
86% had multiple chronic diseases
Higher levels of anaemia
Mental illness higher
Large percentage die in hospital in the 12 months after the “reference” admission
3 models1. The “real time” algorithm” uses “real time” data to
identify level of risk of re-hospitalisation for patients hospitalised for “reference” conditions before they are discharged – requires historic data on hospitalisation as well as daily downloading of data from A&E systems
2. The “monthly” algorithm is designed to be run each month and is based on historic data as well as monthly admission data from NWCS or SUS
3. The “annual” algorithm identifies patients who have been admitted within the year and who are at risk of a subsequent admission in the next 12 months – uses historic NWCS data
Next step – implementing effective interventions
Flexible and match particular needs of each patient
Non-intrusive
Cost-effective
Co-ordinates medical care, social care and community resources
Over to you….
Experiences of using the algorithm
Lessons
Problems
Pitfalls
Advantages